Some tweaks to accommodate relocating development to loca

Co-authored-by: Copilot App <[email protected]>

.env.example

@@ -1,8 +0,0 @@
-PROVIDER=openrouter
-OPENROUTER_API_KEY=sk-or-...
-# PROVIDER_MODEL=          # optional: OpenRouter adapter supplies default
-# OPENROUTER_HTTP_REFERER=https://example.com
-# OPENROUTER_APP_TITLE=Historical Transcription MVP
-# DATABASE_URL=sqlite:///./transcription.db
-# UPLOAD_DIR=./uploads
-# PROMPT_DIR=./prompts

.env.production.example

@@ -0,0 +1,67 @@
+# Production environment example for docker-compose.production.yml
+
+# --- NiceGUI Server ---
+HOST=0.0.0.0
+PORT=8000
+LOG_LEVEL=info
+RELOAD=false
+ENVIRONMENT=production
+# TRANSCRIPTION_COMMIT=
+RUN_EMBEDDED_WORKER=false
+LOG_DIR=/app/data/logs
+LOG_FILE_NAME=transcription.log
+LOG_FILE_MAX_BYTES=10485760
+LOG_FILE_BACKUP_COUNT=5
+
+# --- AI provider ---
+PROVIDER=openrouter
+OPENROUTER_API_KEY=replace-with-real-key
+PROVIDER_MODEL=google/gemini-2.5-flash
+# PROVIDER_MODELS=["google/gemini-2.5-flash","anthropic/claude-sonnet-4"]
+# OPENROUTER_HTTP_REFERER=
+# OPENROUTER_APP_TITLE=
+DEFAULT_PROMPT_NAME=transcribe_document.md
+# TRANSCRIPTION_TEMPERATURE=
+# TRANSCRIPTION_TOP_P=
+
+# --- persistence ---
+# Common database settings:
+DATABASE__DRIVER=postgres
+DATABASE__DATABASE=transcription
+DATABASE__USER=transcription
+DATABASE__PASSWORD=replace-with-strong-password
+BOOTSTRAP_SCHEMA_ON_STARTUP=false
+
+# SQLite-specific settings:
+# DATABASE__PATH=./data/transcription.db
+# SQLITE_CHECK_SAME_THREAD=false
+
+# Postgres-specific settings:
+DATABASE__HOST=postgres
+DATABASE__PORT=5432
+
+# --- filesystem paths ---
+UPLOAD_DIR=/app/uploads
+PROMPT_DIR=/app/prompts
+# --- backup workflow helpers (not Runtime Settings model fields) ---
+BACKUP_DIR=/backup
+BACKUP_RETENTION_DAYS=14
+
+# --- worker reliability ---
+WORKER_MAX_RETRIES=0
+WORKER_PROVIDER_TIMEOUT_SECONDS=30.0
+WORKER_STALE_JOB_SECONDS=90.0
+WORKER_RETRY_BACKOFF_SECONDS=1.0
+WORKER_SHUTDOWN_GRACE_SECONDS=5.0
+WORKER_POLL_INTERVAL_SECONDS=1.0
+WORKER_MIN_TRANSCRIPTION_CHARS=0
+WORKER_MIN_TRANSCRIPTION_LINES=0
+WORKER_FAIL_ON_FINISH_REASON_LENGTH=false
+
+# --- cloudflare tunnel ---
+# Required for token-based tunnel startup.
+CLOUDFLARE_TUNNEL_TOKEN=replace-with-cloudflare-tunnel-token
+
+# --- deployment wiring helpers ---
+# Runtime Settings writes target this file path inside the app container.
+RUNTIME_SETTINGS_ENV_FILE=/app/.env.production

.env.production.postgres

@@ -0,0 +1,67 @@
+# Production environment
+
+# --- NiceGUI Server ---
+HOST=0.0.0.0
+PORT=8000
+LOG_LEVEL=info
+RELOAD=false
+ENVIRONMENT=production
+# TRANSCRIPTION_COMMIT=
+RUN_EMBEDDED_WORKER=false
+LOG_DIR=/app/data/logs
+LOG_FILE_NAME=transcription.log
+LOG_FILE_MAX_BYTES=10485760
+LOG_FILE_BACKUP_COUNT=5
+
+# --- AI provider ---
+PROVIDER=openrouter
+OPENROUTER_API_KEY=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
+PROVIDER_MODEL=google/gemini-2.5-flash
+PROVIDER_MODELS=["google/gemini-2.5-pro","google/gemini-2.5-flash","anthropic/claude-opus-5","anthropic/claude-sonnet-4","openai/gpt-5.6","openai/gpt-4o"]
+# OPENROUTER_HTTP_REFERER=
+# OPENROUTER_APP_TITLE=
+DEFAULT_PROMPT_NAME=transcribe_document.md
+# TRANSCRIPTION_TEMPERATURE=
+# TRANSCRIPTION_TOP_P=
+
+# --- persistence ---
+# Common database settings:
+DATABASE__DRIVER=postgres
+DATABASE__DATABASE=transcription
+DATABASE__USER=transcription
+DATABASE__PASSWORD=<password>
+BOOTSTRAP_SCHEMA_ON_STARTUP=false
+
+# SQLite-specific settings:
+# DATABASE__PATH=./data/transcription.db
+# SQLITE_CHECK_SAME_THREAD=false
+
+# Postgres-specific settings:
+DATABASE__HOST=postgres
+DATABASE__PORT=5432
+
+# --- filesystem paths ---
+UPLOAD_DIR=/app/uploads
+PROMPT_DIR=/app/prompts
+# --- backup workflow helpers (not Runtime Settings model fields) ---
+BACKUP_DIR=/backup
+BACKUP_RETENTION_DAYS=14
+
+# --- worker reliability ---
+WORKER_MAX_RETRIES=0
+WORKER_PROVIDER_TIMEOUT_SECONDS=30.0
+WORKER_STALE_JOB_SECONDS=30.0
+WORKER_RETRY_BACKOFF_SECONDS=1.0
+WORKER_SHUTDOWN_GRACE_SECONDS=5.0
+WORKER_POLL_INTERVAL_SECONDS=1.0
+WORKER_MIN_TRANSCRIPTION_CHARS=0
+WORKER_MIN_TRANSCRIPTION_LINES=0
+WORKER_FAIL_ON_FINISH_REASON_LENGTH=false
+
+# --- cloudflare tunnel ---
+# Required for token-based tunnel startup.
+CLOUDFLARE_TUNNEL_TOKEN=CLOUDFLARE_TUNNEL_TOKEN=eyJhIjoiYTRhNjM0NzNhNzBiZjhhYmY3OWUyNjE4ZTcyNjgwZmMiLCJ0IjoiZWY1MjFkNWItYzY1ZS00ZGFmLTlmYTMtMzQyOGYzMGUyMDY4IiwicyI6IlpHWm1NVGRsTVRVdFpEYzNaaTAwWkRJeUxXRmhPRFV0TmpKallXRmhPRFJrWXpSaSJ9
+
+# --- deployment wiring helpers ---
+# Runtime Settings writes target this file path inside the app container.
+RUNTIME_SETTINGS_ENV_FILE=/app/.env.production

.env.production.sqlite

@@ -0,0 +1,67 @@
+# Production environment
+
+# --- NiceGUI Server ---
+HOST=0.0.0.0
+PORT=8000
+LOG_LEVEL=info
+RELOAD=false
+ENVIRONMENT=production
+# TRANSCRIPTION_COMMIT=
+RUN_EMBEDDED_WORKER=false
+LOG_DIR=/app/data/logs
+LOG_FILE_NAME=transcription.log
+LOG_FILE_MAX_BYTES=10485760
+LOG_FILE_BACKUP_COUNT=5
+
+# --- AI provider ---
+PROVIDER=openrouter
+OPENROUTER_API_KEY=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
+PROVIDER_MODEL=google/gemini-2.5-flash
+PROVIDER_MODELS=["google/gemini-2.5-pro","google/gemini-2.5-flash","anthropic/claude-opus-5","anthropic/claude-sonnet-4","openai/gpt-5.6","openai/gpt-4o"]
+# OPENROUTER_HTTP_REFERER=
+# OPENROUTER_APP_TITLE=
+DEFAULT_PROMPT_NAME=transcribe_document.md
+# TRANSCRIPTION_TEMPERATURE=
+# TRANSCRIPTION_TOP_P=
+
+# --- persistence ---
+# Common database settings:
+DATABASE__DRIVER=sqlite
+# DATABASE__DATABASE=transcription
+# DATABASE__USER=transcription-local
+# DATABASE__PASSWORD=My!3sons
+# BOOTSTRAP_SCHEMA_ON_STARTUP=false
+
+# SQLite-specific settings:
+DATABASE__PATH=./data-local/transcription-local.db
+SQLITE_CHECK_SAME_THREAD=false
+
+# Postgres-specific settings:
+# DATABASE__HOST=postgres
+# DATABASE__PORT=5432
+
+# --- filesystem paths ---
+UPLOAD_DIR=./data-local
+PROMPT_DIR=/data/prompts
+# --- backup workflow helpers (not Runtime Settings model fields) ---
+BACKUP_DIR=/backup
+BACKUP_RETENTION_DAYS=14
+
+# --- worker reliability ---
+WORKER_MAX_RETRIES=0
+WORKER_PROVIDER_TIMEOUT_SECONDS=30.0
+WORKER_STALE_JOB_SECONDS=30.0
+WORKER_RETRY_BACKOFF_SECONDS=1.0
+WORKER_SHUTDOWN_GRACE_SECONDS=5.0
+WORKER_POLL_INTERVAL_SECONDS=1.0
+WORKER_MIN_TRANSCRIPTION_CHARS=0
+WORKER_MIN_TRANSCRIPTION_LINES=0
+WORKER_FAIL_ON_FINISH_REASON_LENGTH=false
+
+# --- cloudflare tunnel ---
+# Required for token-based tunnel startup.
+CLOUDFLARE_TUNNEL_TOKEN=CLOUDFLARE_TUNNEL_TOKEN=eyJhIjoiYTRhNjM0NzNhNzBiZjhhYmY3OWUyNjE4ZTcyNjgwZmMiLCJ0IjoiZWY1MjFkNWItYzY1ZS00ZGFmLTlmYTMtMzQyOGYzMGUyMDY4IiwicyI6IlpHWm1NVGRsTVRVdFpEYzNaaTAwWkRJeUxXRmhPRFV0TmpKallXRmhPRFJrWXpSaSJ9
+
+# --- deployment wiring helpers ---
+# Runtime Settings writes target this file path inside the app container.
+RUNTIME_SETTINGS_ENV_FILE=/app/.env.production

.github/agents/python-reviewer.agent.md

@@ -0,0 +1,23 @@
+---
+name: Python Architect Reviewer
+description: Evidence-based senior architect reviewer for FastAPI, NiceGUI, and SQLModel codebases.
+skills:
+  - python-code-reviewer
+---
+
+# Python Architect Reviewer
+
+You are a Senior Python Architect performing an evidence-based, read-only code review.
+
+> No `tools:` allowlist is declared here on purpose. Tool identifiers differ between the runtimes
+> this agent is invoked from, so a hard-coded list silently under-tools the agent in one of them.
+> Read-only discipline is enforced by the **Read-Only Scope** rule below, not by the frontmatter.
+
+## Operating Principles
+
+- **Stack Context:** Python 3.12+, FastAPI, NiceGUI, SQLModel, SQLAlchemy (SQLite/PostgreSQL), Pydantic V2, asyncio workers, and OpenRouter adapters.
+- **Evidence-Based:** Always inspect real files. Every finding must reference concrete file paths and line numbers (e.g., `app/services/worker.py:45-78`). Do not speculate.
+- **Tool Verification:** This is a `uv` project; the toolchain is not on `PATH`. Verify with `uv run ruff check .`, `uv run ty check`, and `uv run pytest -q -m "not external"`, and record the exact commands and outcomes. Never report a lint, type, or test claim you did not run.
+- **Verify Recommendations, Not Just Findings:** Before recommending a change to a shared symbol, enumerate its consumers and confirm the fix is safe for each. See the skill's consumer-tracing step and `Blast Radius` field.
+- **Skill Is Canonical:** The `python-code-reviewer` skill defines the review workflow, deterministic checks, severity and reachability rubrics, report location, and report template. Follow it exactly. Where this file and the skill disagree, the skill wins — do not restate its specifics here.
+- **Read-Only Scope:** Do not modify source, tests, docs, instructions, or configuration. The review report is the only artifact you produce.

.github/instructions/documentation-sync.instructions.md

@@ -0,0 +1,41 @@
+---
+description: Require documentation updates whenever code changes alter contracts, behavior, or scope.
+applyTo: 'src/transcription/**/*.py'
+---
+
+# Documentation Sync Requirements
+
+Keep docs in sync in the same change whenever implementation alters a documented contract, behavior, or roadmap decision.
+
+Documentation targets below always refer to the **current** baseline. `docs/index.md` states which
+baseline that is; resolve any version-specific document from there. Never cite a superseded version
+tree by name in this file or in the docs you update — retired revision trees are not authority, and
+`tests/test_meta_contract_guards.py` fails active contract files that route authority through them.
+
+## Update documentation when any of these change
+
+1. **Schema/Data contract**
+   - Models, fields, enums, constraints, indexes, relationships, loading semantics.
+   - **Required doc update:** `docs/schema.md`.
+
+2. **Configuration contract**
+   - `Settings` keys, defaults, required/optional environment values.
+   - **Required doc update:** `.env.production.example` and any directly related setup docs.
+
+3. **User-visible UI behavior**
+   - Page flow, routes, button/action behavior, labels, status wording, empty/error states.
+   - **Required doc update:** relevant `docs/ui/pages/*.md` docs and feature docs when applicable.
+
+4. **Error handling semantics**
+   - Error categories, retry behavior, envelope structure, translation boundaries.
+   - **Required doc update:** `docs/error_handling.md` and `docs/invariant/error_handling.md`.
+
+5. **Roadmap/scope decisions**
+   - Version targets, sequencing, deferrals, and accepted alternatives.
+   - **Required doc update:** `docs/roadmap_plan.md`, plus any backlog or feature document for the
+     current baseline. Locate it through `docs/index.md` rather than assuming a version-named path.
+
+## Working rule
+
+If none of the categories above changed, documentation edits are optional.
+If any category changed, update docs in the same PR/change set rather than deferring.

.github/instructions/error-handling.instructions.md

@@ -0,0 +1,121 @@
+---
+description: Cross-cutting error handling rules for services, API, and UI.
+applyTo: 'src/transcription/**/*.py'
+---
+
+# Error Handling (Cross-cutting)
+
+Primary references:
+
+- `docs/error_handling.md`
+- `docs/invariant/error_handling.md`
+- `docs/requirements.md`
+
+## Taxonomy and Categories
+
+Use category-driven semantics aligned to canonical policy:
+
+- `validation`
+- `not_found`
+- `conflict`
+- `external`
+- `timeout`
+- `internal`
+
+Do not invent ad hoc categories in user/API-facing envelopes unless canonical docs are updated.
+
+Runtime/internal categories may be more specific for diagnostics and persistence, but they must map
+deterministically to the canonical envelope categories through the centralized mapper in
+`transcription.errors.canonical_error_category`.
+
+Current internal categories:
+
+- `validation_error`
+- `user_input_error`
+- `not_found_error`
+- `conflict_error`
+- `external_provider_error`
+- `external_timeout_error`
+- `processing_error`
+- `infrastructure_transient_error`
+- `infrastructure_persistent_error`
+- `internal_unexpected_error`
+
+Required internal -> canonical mapping:
+
+- `validation_error`, `user_input_error` -> `validation`
+- `not_found_error` -> `not_found`
+- `conflict_error` -> `conflict`
+- `external_provider_error` -> `external`
+- `external_timeout_error`, `infrastructure_transient_error` -> `timeout`
+- `processing_error`, `infrastructure_persistent_error`, `internal_unexpected_error` -> `internal`
+
+## Translation Boundaries
+
+- **Provider/adapters:** raise provider/domain exceptions; do not emit UI text.
+- **Services:** map raw exceptions into internal categories and preserve causal chain (`raise ... from ...`).
+- **UI/API:** map internal category -> canonical envelope category and emit user-safe, actionable messages.
+
+## Retry Rules
+
+- No auto-retry for `validation`, `not_found`, `conflict`.
+- `external`/`timeout` may be retried when operation semantics are safe.
+- Preserve each retry as new evidence where applicable (no history rewrite).
+
+## Job/Page Failure Semantics
+
+- Page-level (`JobSource`): `pending`, `transcribed`, `failed`, `cancelled`.
+- Job terminals: `transcribed`, `partial_success`, `failed`.
+- Cancellation must keep job-level and page-level semantics explicit and consistent.
+- Do not emit legacy terminal state language such as `completed` in active user/API lifecycle contracts.
+
+## User-Safe Messaging
+
+- Never leak stack traces, credentials, auth headers, or local filesystem paths in user-facing output.
+- Include actionable remediation guidance aligned to category.
+- Keep envelope structure consistent across API endpoints.
+
+### `AppError.message` vs `AppError.detail`
+
+`AppError` carries two texts with different audiences, and they must not be collapsed. Getting this
+wrong has already caused a real defect in this repository, in both directions.
+
+| Attribute | Audience | Reaches | Rule |
+| :--- | :--- | :--- | :--- |
+| `message` | User and API clients | `ErrorEnvelope.message`, UI notifications | Stays generic. Never embed exception text, provider payloads, or filesystem paths. |
+| `detail` | Internal only | Logs, and `format_error_detail` -> `ExecutionAttempt.error_detail` and `MaintenanceRun.error_detail` | Carries the root cause. Never rendered to users or serialized into an envelope without a sanitizing projection. |
+
+- Putting root-cause data in `message` leaks infrastructure detail to users.
+- Omitting it from `detail` silently degrades the provenance record this system exists to preserve —
+  a failed attempt whose `error_detail` says nothing is an attempt that cannot be diagnosed later.
+- Any render boundary that displays persisted `error_detail` must apply the same no-local-path rule
+  as `message`: sanitize machine-local absolute paths before the text becomes user-visible.
+- When you raise from a caught exception, populate **both**: a generic `message` and a `detail`
+  carrying `type(exc).__name__` and the exception text, with `raise ... from exc`.
+- `detail` is optional (`None`). A read path that assumes it is populated must handle its absence.
+- Before changing either attribute, or any helper that formats them, enumerate every consumer —
+  evidence writes, maintenance runs, logging, API envelopes, and UI presentation all read these
+  fields, and tests assert on the persisted text.
+
+Canonical definitions live in `src/transcription/errors.py`; see also `docs/error_handling.md`.
+
+## Logging and Diagnostics
+
+- Log operation identifiers and error IDs where available.
+- Preserve category + cause-chain context.
+- Distinguish no-response timeout/network failures from returned provider error responses.
+
+## Guardrails
+
+- No broad catch-and-swallow patterns.
+- No success-shaped fallback values after exceptions.
+- Category mapping must remain deterministic and testable.
+
+## Contract Sync Rule
+
+If taxonomy, retries, or envelope semantics change:
+
+1. Update canonical docs (`docs/error_handling.md`, and invariant docs if needed).
+2. Update tests in the same change.
+3. Update related instruction/skill references.
+4. If change affects persisted status/category fields, update `docs/schema.md` when applicable.

.github/instructions/providers.instructions.md

@@ -0,0 +1,122 @@
+---
+description: Provider adapter rules for evidence capture, secret safety, and client lifecycle.
+applyTo: 'src/transcription/providers/**/*.py'
+---
+
+# Provider Adapters
+
+Primary references:
+
+- `docs/invariant/ai_evidence_and_provenance.md` (canonical; provider adapters own provider-boundary evidence capture)
+- `docs/architecture.md`
+- `docs/schema.md`
+
+The provider layer is where an external API becomes application data. It is also the only place
+that can capture what actually crossed the wire — once a response reaches a service, the evidence
+it did not preserve is gone permanently. Treat capture correctness as the primary job of this
+layer and text extraction as secondary.
+
+## Layer Boundary
+
+- Adapters may depend on `transcription.config`, `transcription.providers.*`, the HTTP client, and
+  the provider SDK. They must not import `services`, `db`, `ui`, or `api`.
+- Provider specifics — headers, model slugs, payload shapes, SDK types, error classes — stop here.
+  Callers receive `TranscriptionResult` and `ProviderError` subclasses only.
+- Adapters raise provider/domain exceptions. They must not emit user-facing text, notifications,
+  or remediation wording; that translation belongs to services and UI. See
+  [error-handling instructions](./error-handling.instructions.md).
+- Adapters do not persist. They return evidence; services decide what is written and when.
+
+Enforced by `tests/test_provider_boundaries.py`.
+
+## Contract Surface
+
+- Every adapter satisfies the `TranscriptionProvider` protocol in `base.py`. Failed-call evidence is
+  returned through the caller-owned `ProviderCallEvidence` sink passed to `transcribe()`, so
+  evidence stays scoped to one invocation instead of living on mutable adapter instance state.
+- `TranscriptionResult`, `RequestManifest`, and `TransportEvidence` are `extra="forbid"` and frozen.
+  Add a field to the contract rather than smuggling data through an untyped dict.
+- Evidence contracts in `evidence.py` are versioned (`schema_name` + `schema_version`). A change to
+  the meaning or shape of a captured field requires a version bump, not a silent redefinition —
+  stored evidence must keep its original meaning.
+
+## Transport Evidence
+
+The rules below implement `docs/invariant/ai_evidence_and_provenance.md` §3.4-3.5. That document
+wins if this file drifts from it.
+
+- Capture the response body **at the HTTP boundary, before SDK parsing**, so fields the SDK does
+  not model are not lost. `_CapturingAsyncClient` exists for this; do not replace it with a
+  post-parse `model_dump()` and call the result transport evidence.
+- Reset per-call capture state at the start of every call. Without it, a connection failure can
+  attach the *previous* call's response as evidence for this one. Guarded by
+  `tests/test_evidence_provenance.py::test_openrouter_does_not_reuse_prior_response_on_connection_failure`.
+- Keep transport capture scoped to the call, not the adapter instance. Concurrent `transcribe()`
+  calls on one adapter must not be able to overwrite each other's response evidence.
+- Handle the streamed-body case (`httpx.ResponseNotRead`) rather than assuming `response.content`
+  is always available.
+- When no response arrives — timeout, DNS, connection reset — emit
+  `TransportEvidence(response_received=False)`. Absence of a response is itself evidence and must
+  be explicit, never an empty body or a missing record.
+- Preserve safe response evidence for **unsuccessful** calls too, whenever a response was received.
+- Never relabel an SDK snapshot or normalized metadata as transport evidence, and never backfill
+  it into an execution that predates capture.
+
+## Secret Safety
+
+- Persist response headers only through `filter_safe_response_headers` and the
+  `SAFE_RESPONSE_HEADERS` allowlist. Allowlist, never denylist: capture-then-redact is prohibited,
+  because an unknown header is unsafe by default.
+- Adding a header to the allowlist is a deliberate evidence decision. Confirm it carries no
+  credential, cookie, or session material, and state why it is needed for correlation, content
+  interpretation, rate-limit diagnosis, or audit.
+- API keys, `Authorization`, and cookies must never appear in a manifest, evidence record, log
+  line, or exception message.
+- The request manifest references source content by identity (digest, size, media type, page).
+  Do not duplicate base64 source bytes into it — `_replace_embedded_media` exists for this.
+
+## Execution Specification
+
+The manifest must let a reader reconstruct what was asked, per invariant §3.3:
+
+- Provider, requested model, full effective prompt text, and prompt digest.
+- Every explicitly supplied parameter, and — separately — which optional parameters were
+  **omitted**. Omission is not the same as a null value or an assumed provider default; the
+  `optional_parameter_states` distinction between `omitted`, `null`, and `value` is deliberate.
+- Timeout budget, retry policy, source reference, and `SoftwareContext` versions.
+- Manifest digests use `canonical_json_bytes`. Do not hash a plain `json.dumps()`; key order and
+  separators must stay deterministic or digests become uncomparable.
+
+## Client Lifecycle and Async Safety
+
+- Reuse one pooled `AsyncClient` per adapter instance; do not construct a client per request.
+- Accept an injected client so tests can drive the adapter without network access.
+- Derive timeouts from `Settings` (`worker_provider_timeout_seconds`) rather than hard-coding, and
+  keep the client timeout aligned with the configured budget so the SDK cannot expire first and
+  hide the real failure.
+- Implement `aclose()` and release pooled resources. An adapter that creates a client owns closing
+  it; one given a client must not close a caller-owned resource it did not create.
+- Never block the event loop. Offload CPU-bound work (hashing large payloads, image encoding) with
+  `asyncio.to_thread`.
+- Propagate `asyncio.CancelledError` untouched — do not convert cancellation into a provider error.
+
+## Failure Handling
+
+- Raise `ProviderAuthError` for authentication, `ProviderResponseError` for malformed or unusable
+  responses, and `ProviderError` otherwise.
+- Always attach `request_manifest`, `transport_evidence`, and an accurate `failure_phase` to raised
+  errors. `failure_phase` must distinguish a received-but-failed response from a call that never
+  reached the provider.
+- Validate responses with Pydantic rather than indexing into raw dicts.
+- Invalid *optional* metadata (for example unparsable token counts) must not discard an otherwise
+  valid transcript. Degrade the metadata, not the result.
+
+## Contract Sync Rule
+
+If capture behavior, evidence schema, or the header allowlist changes:
+
+1. Update `docs/invariant/ai_evidence_and_provenance.md` only if the durable preservation contract
+   itself is changing — that revision is deliberate and reviewed, not incidental.
+2. Update `docs/schema.md` when persisted evidence fields change.
+3. Update or add tests in the same change (`tests/providers/`, `tests/test_evidence_provenance.py`).
+4. Bump the affected evidence `schema_version` when a field's meaning changes.

.github/instructions/services.instructions.md

@@ -7,29 +7,135 @@ applyTo: 'src/transcription/services/*.py'
 
 ## Structure
 
-- Project core data models defined in [models](../../src/transcription/models.py)
-- 1 service class per data model
-- Only services directly interact with the database, and only through async methods
-- Services are completely independent of one another. Any operation that needs to use more than a single service, which is most of them, needs to have a separate orchestration function.
+- Project core data models are defined in [models](../../src/transcription/db/models.py)
+- One service class per **aggregate**, not per table. An aggregate is a root model plus
+  the models that have no independent lifecycle of their own. `DocumentType` has no
+  meaning without `Document`, so it belongs to `DocumentService`; it does not get its
+  own service. Splitting per table produces services that must reach across each other
+  for every real operation, which is what line 13 forbids.
+- Only services interact with the database, and only through async methods.
+- **A service module must not import another service module.** This is enforced by
+  [test_service_boundaries](../../tests/test_service_boundaries.py). Shared types go in a
+  neutral module that defines no service class (see [errors](../../src/transcription/services/errors.py)).
+- Not every module in this package is a service. Modules fall into three kinds:
+  - **Aggregate services** own models and define a `*Service` class: `documents.py`, `sources.py`,
+    `jobs.py`, `people.py`, `photos.py`, `maintenance.py`, and `evidence.py` (read/projection only,
+    owns nothing).
+  - **Orchestration modules** define no service class and compose writes across aggregates:
+    `store.py`, `workflows.py`. They are the sanctioned place to create or delete rows owned by more
+    than one service — see [Service Composition](#service-composition).
+  - **Shared infrastructure and free-function helpers** are exempt from the service rules below:
+    `base.py` (`ServiceBase`), `registry.py` (`RegistryService`, a generic base for lookup tables —
+    not an aggregate owner itself), `unit_of_work.py`, `errors.py`, `normalization.py`, `prompts.py`,
+    `quality.py`, `media_storage.py`, `source_media.py`. `__init__.py` exposes `ServiceBundle`.
+- Cross-cutting error behavior must follow
+  [error-handling instructions](./error-handling.instructions.md).
+
+## Model Ownership
+
+Every model has exactly one owning service. The owner defines that model's invariants and
+is the only service that may **create or delete** its rows.
+
+| Model | Owner |
+| --- | --- |
+| `Document`, `DocumentType`, `DocumentTag` | `DocumentService` |
+| `Source`, `JobSource` | `SourceService` |
+| `Job` | `JobService` |
+| `Person`, `PersonRole`, `DocumentPerson`, `PersonTag` | `PeopleService` |
+| `GenealogyPerson`, `GenealogyFamily`, `GenealogyFamilyChild`, `GenealogyCitation` | `MaintenanceService` |
+| `Photo` | `PhotosService` |
+| `MaintenanceRun` | `MaintenanceService` |
+| `ExecutionAttempt` | `SourceService` |
+| `Tag` | shared — see below |
+
+Keep this table complete: every table in `src/transcription/db/models.py` appears exactly once,
+except `Tag`. When you add a model, add its owner here in the same change.
+
+### `Tag` is deliberately shared
+
+`Tag` is one table reached through two `RegistryService[Tag]` facades that differ only in the
+reference model they count usage through: `TagRegistry` (`documents.py`, via `DocumentTag`) and
+`PersonTagRegistry` (`people.py`, via `PersonTag`). Both create and delete `Tag` rows through the
+generic registry. This is the single sanctioned exception to one-owner-per-model — do not "fix" it by
+assigning `Tag` to one service, because the other facade would then be creating rows it does not own.
+Any change to `Tag` semantics, labels, or normalization must be validated against **both** facades
+and the junction table each one counts.
+
+`DocumentTag` and `PersonTag` follow the junction rule below: each is created and deleted only by the
+service on its own side.
+
+### Registries
+
+`DocumentTypeRegistry`, `TagRegistry`, `PersonRoleRegistry`, and `PersonTagRegistry` are
+`RegistryService` subclasses, not independent services. A registry belongs to the aggregate service
+whose module declares it and shares that service's ownership. Registry CRUD uses `<operation>_entry`
+naming (see [CRUD Methods](#crud-methods)).
+
+### Junction tables
+
+A junction table is owned by the service that **creates and deletes its rows** — its
+lifecycle owner. The service on the other side may read through the junction (via
+`selectinload`) but must not create rows in it.
+
+- `document_person` -> `PeopleService`. Every write is there; `DocumentService` only
+  eager-loads through it.
+- `job_source` -> `SourceService`, which creates the row, records each page's outcome,
+  and deletes it.
+
+Two consequences follow, and both are deliberate:
+
+- **Cascade deletion is not a violation.** A service deleting the aggregate root it owns
+  may delete rows referencing that root which cannot outlive it
+  (`JobService.delete_job_with_guardrails` deletes the job's `job_source` rows).
+- **Evidence deletion is an explicit workflow, not a runtime path.** `JobService.delete_job_and_evidence`
+  deletes `ExecutionAttempt` rows owned by `SourceService`. That is sanctioned because it is the
+  named retention workflow that `delete_job_with_guardrails` refuses to perform implicitly — that
+  method *blocks* deletion when attempts exist. Append-only means runtime code never rewrites or
+  removes history to represent a new outcome; it does not forbid a deliberate, operator-invoked
+  retention operation. Do not add a second path that deletes attempts.
+- **Ownership governs creation and deletion, not every state transition.** `job_source` is
+  both a link and the transcription work queue. `JobService.cancel_job` and
+  `resubmit_failed_sources` transition `job_source.status` across a whole job, because that
+  transition is a Job lifecycle event, not a per-page outcome. They create and delete
+  nothing.
+
+`EvidenceService` is read-focused and projection-focused. It may coordinate selection
+flows, but append-only attempt creation remains in `SourceService` write paths.
+
+If a new operation cannot be expressed within one owner, it belongs in an orchestration
+module, not in a cross-service import.
 
 ## Error Handling
 
-- Service-specific errors defined at the top of the respective module and inherit from `AppError`
-- Use a context manager for large `try/except` blocks like in [transcription](../../src/transcription/services/transcription.py)
+- Errors used by a single service are defined at the top of that module and inherit from `AppError`.
+- Errors shared by more than one service go in [errors](../../src/transcription/services/errors.py),
+  which defines no service class and is therefore importable by any of them.
+- Use a context manager for large `try/except` blocks, like `handle_transcription_errors` in
+  [sources](../../src/transcription/services/sources.py).
+- Category mapping, retry behavior, and translation boundaries are defined in
+  [error-handling instructions](./error-handling.instructions.md).
+- Service-edge exception translation must be deterministic: map to canonical categories and preserve clear provider->service->API/UI boundaries.
 
 ## Checklist
 
 - [ ] Uses `ServiceBase` for common logic
-- [ ] CRUD methods created at the top
-- [ ] Session kwarg for `AsyncSession` to pass in a session object to each method
-- [ ] Services use `self._session_scope` in their methods to pass the session thru.
-    - Multiple operations on the same object(s) require sharing a session between all the methods used.
+- [ ] Session kwarg for `AsyncSession` to pass a session object into each method
+- [ ] Services use `self._session_scope` in their methods to pass the session through
+  - Multiple operations on the same object(s) require sharing a session between all the methods used
+- [ ] Every model the module touches is either owned by it or reached read-only
+- [ ] Evidence writes preserve append-only semantics
 
 ## CRUD Methods
 
-- Create, read, update, and delete, created in that order
-- Name format `<operation>_<model >`, for example `create_document` or `update_job`
-- All services must define these 4 methods first, and in that order
+- Name format `<operation>_<model>`, for example `create_document` or `update_job`.
+- Where a service exposes create/read/update/delete for its root model, define them at the
+  top of the class in that order, before derived reads and workflow helpers.
+- Not every aggregate needs all four. `ExecutionAttempt` is append-only evidence written by
+  `SourceService` workflow-facing methods, so `EvidenceService` deliberately exposes reads and
+  no create or delete.
+  Do not add unused CRUD methods to satisfy symmetry.
+- `RegistryService` is generic across small lookup models and uses `<operation>_entry`
+  naming instead.
 
 ## Transaction Finalization
 
@@ -39,18 +145,9 @@ When a service method accepts an optional `session` kwarg, write methods must us
 - If `session` is provided: the method must **not** commit; it should `flush()` so IDs and FK values are available to the caller's transaction.
 - Use `refresh()` on returned ORM objects when the caller needs DB-populated values (defaults, triggers, merged state).
 
-Recommended helper behavior:
-
-- Inputs: active session object, original `session` arg (or a boolean ownership flag), and an optional list of objects to refresh.
-- Logic: `commit` when service-owned session, `flush` when caller-owned session, then refresh requested objects.
-
-This keeps orchestration functions atomic: they can pass one shared session across multiple services and commit exactly once at the workflow boundary.
-
 ## Workflow Transaction Boundaries
 
-For multi-step job lifecycles (for example queued transcription jobs), orchestration functions must use explicit transaction phases.
-
-Required boundary model:
+For multi-step job lifecycles, orchestration functions must use explicit transaction phases.
 
 - **Transaction A (claim):** transition `JobStatus.QUEUED -> JobStatus.PROCESSING` and commit immediately.
 - Perform provider/network work **outside** database transactions.
@@ -64,14 +161,55 @@ Atomicity rules:
 - Terminal state (`TRANSCRIBED` or `FAILED`) and transcript row changes must succeed or roll back together.
 - Retry persistence (`QUEUED` + retry increment + error detail) must succeed or roll back together.
 
-Separation of concerns:
+### Multi-page batches
 
-- Worker modules should stay lightweight and delegate lifecycle transitions to service/workflow orchestration functions.
-- In `workflows.py`, `process_queued_job` should own one complete attempt lifecycle: `QUEUED -> PROCESSING -> TRANSCRIBED|FAILED`.
-- In `workflows.py`, `advance_job` should coordinate broader status progression around attempts (for example retry scheduling from `FAILED -> QUEUED`).
-- Services should expose session-aware write helpers (flush on caller-owned session) so orchestration controls commit boundaries.
-- Backoff/sleep behavior must run outside transactional scopes.
+These two requirements are in tension for multi-page jobs: each page should be durable as
+soon as its provider call returns, but the last page must commit together with the terminal
+status. `process_queued_job` resolves it by committing every page except the last one
+individually, then deferring the final page's write into `_finalize_batch_outcome` so it
+shares the terminal transaction.
+
+Both paths are shielded against cancellation, so the final page is no less durable than the
+pages before it. Enforced by `tests/integration/test_pipeline_atomicity.py`; per-page
+durability is separately enforced by
+`tests/services/test_workflows_reliability.py::TestWorkflowReliability::test_transcribed_page_is_committed_before_next_provider_call_finishes`.
+
+### Stale-reclaim safety
+
+- `WORKER_STALE_JOB_SECONDS` must remain **greater than** `WORKER_PROVIDER_TIMEOUT_SECONDS`; stale
+  recovery must not be able to fire before one provider call can legitimately finish.
+- Long-running multi-page orchestration must refresh job liveness explicitly between intermediate
+  page commits. Do not rely on incidental row updates or provider metadata writes to keep
+  `Job.date_updated` fresh.
+- Enforced by `tests/test_config.py` and
+  `tests/services/test_workflows_reliability.py::TestWorkflowReliability::test_intermediate_page_commit_advances_job_liveness_timestamp`.
+
+## Contract Alignment
+
+- Treat `docs/` as the active architecture and requirements baseline.
+- Legacy revision trees are out of scope for active implementation decisions and must not be referenced as authoritative service guidance.
+- Treat `src/transcription/db/models.py` as runtime schema ground truth and `docs/schema.md` as the field-accurate contract mirror.
+- `Job.status` success path is `TRANSCRIBED`.
+- `JobSource.status` is queue/projection state only (`PENDING`, `TRANSCRIBED`, `FAILED`, `CANCELLED`).
+- Source ingest may normalize media before persistence; persisted bytes/hash are canonical for processing and provenance.
+- `ExecutionAttempt` is append-only evidence history; do not mutate historical attempt rows in runtime code.
+- `Source.raw_transcription` is a projection, not authoritative history.
+- Service/UI read paths that touch relationships must be eager-loaded for `lazy="raise"` compatibility.
+- If model fields, enums, constraints, indexes, or relationship-loading semantics change, update `docs/schema.md` in the same change.
+- If `Settings` fields or defaults change in `src/transcription/config.py`, update `.env.production.example` in the same change so keys/defaults remain synchronized and no stale settings remain documented.
+
+## Schema Drift and Legacy Compatibility Policy
+
+- Prefer schema migration over startup reconciliation or runtime compatibility paths in service writes.
+- Do not add legacy read/write compatibility code in service workflows by default.
+- If drift is discovered and a migration decision is ambiguous (for example, one-way destructive DDL, uncertain data retention impact, or unknown deployment sequence), pause and ask the user to choose migration vs compatibility before coding.
+- If a temporary compatibility path is explicitly approved, document an expiration/removal plan in the same change.
 
 # Service Composition
 
-Some operations, like uploading a picutre, require modifications to multiple tables, which can be done by composing methods from the service object into a separate function.
+A service method may read across models it does not own, using eager loads from its own
+aggregate root. What it may not do is import another service.
+
+Operations that must **write** models owned by more than one service are composed in an orchestration module
+([store](../../src/transcription/services/store.py),
+[workflows](../../src/transcription/services/workflows.py)).

.github/instructions/tests.instructions.md

@@ -0,0 +1,119 @@
+---
+description: Authoring rules for the test suite, including markers, async discipline, and guard-test design.
+applyTo: 'tests/**/*.py'
+---
+
+# Tests
+
+Primary references:
+
+- `AGENTS.md` (Change Protocol — failing test first)
+- `docs/index.md` and `docs/invariant/*`
+- `.github/skills/test-effectiveness-auditor/skill.md` (periodic audit of this suite)
+
+The suite is not only regression protection here — it is where several architectural rules are
+*defined*. `tests/test_service_boundaries.py`, `tests/test_ui_boundaries.py`,
+`tests/test_provider_boundaries.py`, `tests/test_model_contract_guards.py`, and
+`tests/test_meta_contract_guards.py` are the enforcement layer named in the `AGENTS.md` authority
+order. A weak test in this repository does not merely fail to catch a bug; it can silently repeal a
+documented invariant.
+
+The baseline is green. `uv run pytest -q -m "not external"` must report zero failures and zero
+errors, and there is no tolerated set of known-failing tests.
+
+## Write the Failing Test First
+
+For any behavioral fix, write the test before the fix and confirm it fails *for the reason you
+expect*. A test that passes against the broken code proves nothing, and several defects in this
+repository were subtle enough that a test written afterward would have done exactly that. If the
+new test passes immediately, you have not reproduced the defect yet.
+
+## Runner Configuration
+
+Configured in `pyproject.toml`; do not work around these:
+
+- `--strict-markers` — an unregistered marker is an error. Register new markers in
+  `[tool.pytest.ini_options] markers` with a description rather than inventing one at the call site.
+- `asyncio_mode = "strict"` — every async test needs an explicit `@pytest.mark.asyncio`, and async
+  fixtures use `@pytest_asyncio.fixture`. There is no implicit promotion.
+- `filterwarnings = ["error:coroutine .* was never awaited:RuntimeWarning"]` — an un-awaited
+  coroutine is an error, not a warning. This usually means a mock replaced an async callable with a
+  sync one, or an `await` was dropped. Fix the call; never silence the warning.
+
+## Markers and Layout
+
+- `unit` — pure logic, no framework or database.
+- `integration` — touches framework, database, or multi-component contracts.
+- `external` — calls live services; slow and credential-dependent.
+
+`external` tests must also carry their own `skipif` so the suite stays green without credentials
+(see `tests/services/test_transcription_external.py`). Local and documented runs use
+`-m "not external"`; CI intentionally runs unfiltered, which is equivalent because those tests skip
+themselves. Never let an unmarked test reach the network.
+
+Place tests by the layer under test: `tests/services/`, `tests/ui/`, `tests/api/`,
+`tests/providers/`, `tests/integration/`, with cross-cutting guards at the top level.
+
+## Fixtures and Isolation
+
+- Prefer the shared fixtures in `tests/conftest.py` (`default_settings`, `async_session`,
+  `default_session_factory`, and the per-aggregate service fixtures) over building settings or
+  engines by hand.
+- `Settings` is isolated suite-wide by the session-scoped autouse fixture in `conftest.py`, because
+  `env_file` resolves against the working directory. Tests that need env-file loading pass
+  `_env_file=` explicitly; tests asserting declared defaults need nothing. Do not reintroduce
+  reliance on a developer's local env file. Guarded by `tests/test_config_isolation.py`.
+- Database fixtures refuse to run against anything but the per-test path, and that refusal is
+  deliberate. Never relax it to point a destructive fixture at a real database.
+- Tests must not leave artifacts outside `tmp_path`.
+
+## Assertion Strength
+
+Assert on the domain effect, not on the fact that code ran.
+
+- Prefer persisted state, status transitions, error categories, and evidence records over
+  "no exception raised", "not None", or a bare status code.
+- **Read committed state through a separate session.** Asserting against the same session that
+  performed the write can pass on unflushed in-memory state and prove nothing about durability.
+  This is how the atomicity guarantees in `tests/services/test_workflows_reliability.py` and
+  `tests/integration/test_pipeline_atomicity.py` are made real.
+- Critical paths need negative-path coverage — timeouts, provider failures, validation errors,
+  cancellation. Happy-path-only coverage of a critical module is a gap, not a suite.
+- Avoid count-threshold assertions as a proxy for correctness. A test asserting "at least N items
+  were discovered" passes indefinitely while the thing it was meant to protect rots; assert on a
+  specific known member instead.
+
+## Guard Tests
+
+Structural guards carry extra obligations, because they are cited as proof that a rule holds.
+
+- **Guard the guard.** Every scanning guard needs a companion assertion that the scan actually found
+  something, following the existing `test_*_are_discovered` pattern. A guard that silently scans an
+  empty set passes forever.
+- **Scope must match the claim.** A guard's name and docstring must describe only what it actually
+  verifies. A test covering one function while appearing to enforce a repo-wide rule is worse than
+  no test, because it stops anyone from writing the real one.
+- **Prove non-vacuity by injected fault.** Temporarily introduce the violation, confirm the guard
+  fails with a comprehensible message, then revert. Do this whenever you add or materially change a
+  guard. Revert with an explicit edit if the file has uncommitted changes — `git checkout --` will
+  discard them.
+- **Prefer structural analysis to substring matching.** AST inspection of imports and definitions is
+  resistant to false negatives; a bare-name search across the repository is not, since an unrelated
+  mention anywhere makes dead code look reachable.
+- Failure messages should name the offending file, symbol, and the remedy. These fire for people who
+  did not write the guard.
+- Any new file under `.github/**` must be added to `ACTIVE_CONTRACT_FILES` in
+  `tests/test_meta_contract_guards.py`, or the completeness guard fails by design.
+
+## Redundancy
+
+Duplicate coverage across layers costs runtime and dilutes signal. Pick the canonical layer for a
+behavior — unit for logic, integration for wiring — and let the other layer assert only what is
+unique to it. Retire tests superseded by a stronger guard instead of accumulating both, and record
+deliberate retentions with a rationale rather than leaving them unexplained.
+
+## Contract Sync Rule
+
+When a test encodes or relaxes a documented rule, update the corresponding instruction file or
+`docs/*` page in the same change. When a guard test is the enforcement for a rule stated in
+`AGENTS.md` or an instruction file, cite the test by name there so the link survives refactoring.

.github/instructions/ui.instructions.md

@@ -1,6 +1,78 @@
 ---
-description: Copilot rules for modifying the UI
+description: "Use when modifying the NiceGUI application under src/transcription/ui. Defines ownership and dependency boundaries for UI registration, pages, components, services, persistence, state, and static assets."
 applyTo: 'src/transcription/ui/**/*.py'
 ---
 
-The UI is forbidden from directly touching the database. All interactions should be done using the [services](../../src/transcription/services/)
+# UI Conceptual Boundaries
+
+Keep dependencies flowing in this direction:
+
+`ui/__init__.py` -> `pages` -> `components`
+
+Pages may depend on application services and framework-provided dependencies. Components may depend on smaller components and shared presentation helpers. Services and domain modules must never depend on the UI.
+
+Cross-cutting error behavior must follow
+[error-handling instructions](./error-handling.instructions.md).
+
+## Package Root
+
+- Keep `ui/__init__.py` as the UI composition root: register global assets, register pages, and mount NiceGUI on FastAPI.
+- Do not put feature rendering, service calls, persistence, or route-specific state in the package root.
+
+## Pages
+
+- Pages own route registration and route-level orchestration.
+- Resolve request or application dependencies, call [services](../../src/transcription/services/), adapt returned data for presentation when needed, and coordinate refresh, navigation, and notifications here.
+- Do not query, mutate, commit, or roll back the database from a page. Do not import database engines, sessions, operations, or query-building APIs. Persistence belongs to services or workflow functions.
+- Framework dependency types may cross into page handlers only to construct or invoke services; do not pass sessions or session factories into components.
+- Keep business rules, lifecycle transitions, transaction boundaries, and cross-service workflows out of page callbacks.
+
+## Components
+
+- Components own reusable rendering, widget-local state, input normalization, and presentation-only formatting.
+- Expose user actions through typed callback parameters. The calling page decides which service or workflow runs and what refresh or navigation follows.
+- Do not register routes, resolve request/app state, instantiate services, or access persistence from components.
+- Components may accept ORM models returned by services as read-only snapshots. Only use fields and relationships that the service loaded eagerly; never mutate models, trigger lazy loading, or expose session behavior.
+- A component may compose lower-level components, but it must not import from `pages`.
+
+## Shared UI Infrastructure
+
+- Keep app-wide navigation and layout primitives in `components/app_shell.py`.
+- Keep generic table/event adaptation in `components/table/common.py`; feature-specific columns, row read models, and formatting belong in the feature table module.
+- Keep exception normalization and user-facing error display in `components/error_presenter.py`; preserve `AppError` details and operation identifiers at page/component boundaries.
+- Use `components/media_urls.py` for media URL generation; do not hand-build upload/static paths in page code.
+
+## CSS Assets
+
+- Keep all application CSS in `ui/static/theme.css`; do not add page- or component-specific stylesheets or embed style blocks in Python components.
+- Load `theme.css` once from the composition root with `ui.add_css(..., shared=True)`.
+- Read stylesheet text through `importlib.resources.files(...)` so loading works from installed packages and is independent of the working directory.
+- Centralize CSS reading in one typed helper cached by resource path.
+- Do not encode application behavior in CSS or other static assets.
+
+## State and Side Effects
+
+- Limit component state to ephemeral interaction state such as loading flags, form values, dialogs, and expansion state.
+- Application and worker state must be resolved at the page or application boundary and passed through narrow interfaces.
+- Keep filesystem, network, provider, and worker orchestration behind application services or dedicated adapters.
+
+## Media Route Safety Rules
+
+Two patterns are approved:
+
+1. **Record-validated API routes** for print/export contexts.
+2. **Controlled upload URL resolver** (`components/media_urls.py`) for general UI media.
+
+Prohibited patterns:
+
+- Direct `file://` links or exposing local filesystem paths.
+- Manual URL construction from raw `Path` values in pages/components.
+- User-facing payloads containing local absolute paths.
+
+## Contract Alignment
+
+- Treat `docs/` as the active baseline.
+- Resolve lifecycle and status semantics against `src/transcription/db/models.py` and `docs/schema.md`; do not introduce alternate status labels or implied legacy states in UI behavior.
+- Use status vocabulary exactly as modeled (`queued`, `processing`, `transcribed`, `partial_success`, `failed`; and `pending`, `transcribed`, `failed`, `cancelled`).
+- Print/export media flows must use record-validated routes; direct local filesystem paths are prohibited.
+- If lifecycle wording/behavior changes, update corresponding `docs/ui/pages/*.md` contracts in the same change.

.github/prompts/review-python-architecture.prompt.md

@@ -0,0 +1,23 @@
+---
+name: Review Python Architecture
+description: Run an evidence-based architectural code review using the Python Architect Reviewer agent and python-code-reviewer skill.
+agent: Python Architect Reviewer
+---
+
+# Instructions
+
+Execute a comprehensive, evidence-based code review of the target codebase.
+
+## Target Scope
+- **Review Target:** the repository root, unless the invoker names a narrower path; review that path instead.
+- **Source Root:** `src/`
+- **Docs Root:** `docs/`
+- **Focus Areas:** FastAPI endpoints, NiceGUI components, SQLModel persistence, asyncio workers, Pydantic V2 models, and OpenRouter provider adapters.
+
+## Execution Rules
+1. Map repository layout, dependency manifests, and configuration files from the project root before inspecting modules.
+2. Read real code modules under `src/` (or the specified target path); cite exact file paths and line ranges for every finding.
+3. Validate issues by running `uv run ruff check .`, `uv run ty check`, and `uv run pytest -q -m "not external"`. Nothing is on `PATH` in this `uv` project, so bare `ruff`/`ty`/`pytest` will fail.
+4. Check for duplication, divergent implementations, and extractable helpers.
+5. Format the entire review following the standardized 10-section template defined in the `python-code-reviewer` skill.
+6. Write the final report to `./docs/reviews/<YYYY-MM-DD>-code-review.md`, using today's date. This path is defined by the skill; do not write the report anywhere else.

.github/skills/evidence-provenance-auditor/skill.md

@@ -0,0 +1,80 @@
+---
+name: evidence-provenance-auditor
+description: Deterministic reviewer for transcription evidence/provenance guarantees. Use when changes touch execution attempts, source storage, retries, transport evidence, artifact provenance, or evidence exports.
+---
+
+# Evidence & Provenance Auditor
+
+Perform focused, deterministic audits of evidence integrity and provenance behavior.
+
+## When to Use
+
+- Reviewing changes in:
+  - `src/transcription/services/sources.py`
+  - `src/transcription/services/store.py`
+  - `src/transcription/services/workflows.py`
+  - `src/transcription/services/evidence.py`
+  - `src/transcription/db/models.py`
+- Auditing evidence exports/imports or evidence-display behavior.
+- Verifying no drift from canonical provenance invariants.
+
+## Normative References (must be used)
+
+1. `docs/invariant/ai_evidence_and_provenance.md`
+2. `docs/schema.md`
+3. `docs/requirements.md`
+4. `docs/error_handling.md`
+
+## Deterministic Pass/Fail Checks
+
+### A. Append-only history
+- Every provider call results in a new `ExecutionAttempt`.
+- Runtime paths do not mutate historical attempts to represent new outcomes.
+- Retry behavior appends attempts rather than rewriting prior rows.
+
+### B. Projection vs authority separation
+- `Source.raw_transcription` and preferred pointers are mutable projection surfaces.
+- Attempt rows remain authoritative historical evidence.
+- Candidate promotion updates projection pointers without rewriting history.
+
+### C. Transport evidence semantics
+- Transport evidence is correctly labeled as application-boundary capture.
+- SDK snapshots/normalized metadata are not mislabeled as native upstream payload.
+- No-response timeout/network states are explicit.
+
+### D. Canonical source identity
+- Canonical stored bytes/hash/size are internally consistent.
+- If ingest normalization is applied, code/docs consistently represent resulting canonical identity.
+- Post-ingest derivatives do not overwrite canonical source bytes.
+
+### E. Secret safety
+- No credentials/auth headers/cookies/unrestricted headers persisted.
+- Header persistence uses explicit allowlist semantics.
+
+### F. Route/path safety
+- Print/export source access is record-validated.
+- UI/media path construction does not expose local filesystem paths.
+
+### G. Schema/docs alignment
+- Evidence-related model fields and semantics align with canonical docs.
+- Evidence model changes require same-change doc updates.
+
+### H. Canonical authority boundaries
+- Active guidance resolves against `docs/*` and current instruction files.
+
+## Review Workflow
+
+1. Read normative references first.
+2. Inspect model + service + workflow write paths.
+3. Inspect evidence read/display/export paths.
+4. Report high-confidence findings with concrete path/line evidence.
+5. Classify each finding by invariant family (A-H).
+
+## Output Format
+
+Use this structure:
+
+- Verdict by invariant family (A-H)
+- Findings with `Location`, `Observed Behavior`, `Risk`, `Recommended Fix`
+- Drift table (`Doc claim` vs `Code reality` vs `Action`)
+- Regression guards needed

.github/skills/python-code-reviewer/skill.md

@@ -0,0 +1,276 @@
+---
+name: python-code-reviewer
+description: Perform an evidence-based, senior architect code review for Python codebases using FastAPI, NiceGUI, SQLModel, SQLAlchemy, Pydantic V2, asyncio, and OpenRouter. Use when asked to review Python repositories, perform architectural or code audits, or evaluate code against Python 3.12+ best practices.
+---
+
+# Python Code Reviewer
+
+Perform thorough, evidence-based code reviews for Python projects. Every finding must cite concrete file paths and line ranges, avoid speculation, and include recommended fixes.
+
+## When to Use
+
+- Performing an architectural or code quality review of a Python codebase.
+- Auditing applications using FastAPI, NiceGUI, SQLModel/SQLAlchemy, Pydantic V2, or asyncio workers.
+- Generating structured Markdown review reports in `./docs/reviews`.
+
+## Technical Stack Scope
+
+- **Runtime:** Python 3.12+
+- **Web Application:** FastAPI and NiceGUI
+- **Persistence:** SQLModel, SQLAlchemy (SQLite and PostgreSQL support)
+- **Validation & Settings:** Pydantic V2 and pydantic-settings
+- **Concurrency:** Python asyncio workers
+- **Vision/LLM Integration:** OpenRouter / provider adapters
+- **Image & Print Pipeline:** Pillow-backed media handling and print/export rendering
+- **Quality & Testing:** pytest, pytest-asyncio, Ruff, and ty
+
+NiceGUI is pinned to an exact version (`nicegui==3.13.0` in `pyproject.toml`); API guidance
+must be correct for that release rather than for the latest published version. The exact pin is
+a deliberate release-stability decision recorded in `docs/production-runbook.md` ("Dependency
+upgrade policy") — do not report it as a defect or recommend widening it.
+
+## Review Workflow
+
+1. **Map the Repository First:** Inspect entry points, package layout, configurations, dependency manifests, and any project-specific rule files (`AGENTS.md`, `.github/instructions/`, `.github/skills/`). Project-specific conventions override generic advice.
+2. **Establish Canonical Authority First:** Read architecture/contracts (`docs/*`, `docs/invariant/*`, UI docs) and active instructions/skills before evaluating source behavior.
+3. **Read Representative Modules:** Sample across all layers (routes/pages, UI components, services, workers, persistence, provider adapters, settings, tests) before drawing conclusions.
+4. **Run Drift Analysis:** Compare documented intended behavior versus repository ground truth; identify both implementation drift and undocumented-but-repeatable conventions that should be formalized.
+5. **Run Dead-Code/Orphan Sweep:** Identify candidate orphan modules/functions/classes with zero inbound references, then verify expected exceptions (entrypoints, framework/plugin registration, dynamic imports/reflection, CLI hooks, test-only utilities) before marking as orphaned.
+6. **Assess Boundary and Coupling Health:** Evaluate UI/service/persistence/provider dependency flow, identify circular dependencies, leaky abstractions, and transaction ownership ambiguity.
+7. **Assess Invariant Placement:** For each hard rule, decide whether it belongs in docs (rationale), instructions (active steering), skills (periodic audit procedure), or deterministic tests (enforcement).
+8. **Verify Claims:** This is a `uv` project (`uv.lock`, root `ruff.toml`). Run `uv run ruff check .`, `uv run ty check`, and `uv run pytest -q -m "not external"` rather than guessing, and record the exact commands and their outcomes in the report.
+9. **Validate Recommendations Against Consumers:** A recommendation is a claim about the future and must be verified like any other. Before recommending a change to a shared symbol — a model field, an exception attribute, a helper's return value, a function signature — enumerate **every** consumer of that symbol (`grep` the whole repo, including tests) and confirm the fix is safe for each one. Record the consumers in the finding's **Blast Radius**. A fix that is correct for the path that produced the finding can silently break a second consumer, and evidence/provenance and logging paths are the usual casualties because they read the same fields the UI does.
+10. **Prioritize Hot Paths:** Focus deeply on request handling, database sessions, background workers, and external API calls.
+11. **Enforce Read-Only Safety:** Do not modify code unless explicitly instructed.
+12. **Escalate Provenance Audits:** For evidence/provenance-heavy changes, apply invariant checks from `.github/skills/evidence-provenance-auditor/skill.md` and include pass/fail outcomes in the report.
+13. **Escalate Test-Suite Audits:** When findings touch test coverage, redundancy, or assertion strength, apply `.github/skills/test-effectiveness-auditor/skill.md` and include its outcomes alongside the provenance results.
+
+### Worked example: why step 9 exists
+
+The 2026-08-23 review recommended fixing a filesystem-path leak in
+`classify_unexpected_error` by making `AppError.message` generic and logging the exception
+detail instead. The analysis of the leak was correct, and the fix was implemented as written.
+
+It was wrong. `AppError.message` had a second consumer the review never traced:
+`format_error_detail`, which writes `ExecutionAttempt.error_detail` — a **provenance record**.
+The recommended fix closed a privacy leak by silently stripping root-cause data from the
+evidence history this system exists to preserve. It was caught only because an unrelated
+integration test asserted on the persisted error text.
+
+The correct fix separated the audiences — a user-safe `message` and an internal-only `detail`
+that still reaches evidence and logs. One `grep` for consumers of `.message` during the review
+would have found this. Treat any recommendation that changes a widely-read field as unverified
+until its consumers are enumerated.
+
+## Repo-Specific Deterministic Checks (Transcription)
+When reviewing this repository, always include explicit pass/fail checks for the following.
+Where **Enforced by** reads *unenforced*, recommending a deterministic test is itself a finding.
+
+| # | Check | Enforced by |
+| :-- | :--- | :--- |
+| 1 | **Service boundary rule:** no service-to-service imports | `tests/test_service_boundaries.py` |
+| 2 | **UI boundary rule:** pages/components do not perform persistence access | `tests/test_ui_boundaries.py` |
+| 3 | **Status vocabulary conformance:** `JobStatus`/`JobSourceStatus`/`JobPurpose` usage matches current enums in `src/transcription/db/models.py`; no stringly-typed status literals | `tests/test_model_contract_guards.py` |
+| 4 | **Evidence ownership conformance:** append-only attempt history is preserved and projection writes are not mistaken for history mutation (`src/transcription/services/sources.py`, `src/transcription/services/evidence.py`) | `tests/test_evidence_provenance.py::test_attempts_are_append_only_and_exported_with_integrity` |
+| 5 | **Canonical authority:** findings must resolve against `docs/*` first | `tests/test_meta_contract_guards.py::test_canonical_authority_references_are_present` |
+| 6 | **Schema contract fidelity:** when model/persistence behavior changes, `docs/schema.md` remains field-accurate with `src/transcription/db/models.py` | `tests/test_model_contract_guards.py` (field names, ordering, enum members, table coverage), `tests/test_meta_contract_guards.py` (presence and references) |
+| 7 | **Media boundary conformance:** print/export media is record-validated and UI media URL generation uses controlled resolver paths | `tests/test_media_path_safety.py`, `tests/ui/test_media_urls.py` |
+| 8 | **Eager-loading conformance:** service/UI read paths satisfy `lazy="raise"` expectations | `tests/test_model_contract_guards.py` (declaration-side; documented `noload` exceptions must match `docs/schema.md`) |
+| 9 | **Cross-cutting error conformance:** service/API/UI translation and retry behavior align with `.github/instructions/error-handling.instructions.md` | `tests/test_errors.py`, `tests/api/test_error_responses.py`, `tests/ui/test_error_presenter.py` |
+| 10 | **Orphaned/dead-code conformance:** include a deterministic orphan sweep and report confirmed orphans removed/retained with rationale | `tests/test_orphan_sweep.py` (`KNOWN_ORPHANS` records each retained orphan and its rationale) |
+
+## Core Review Areas
+
+### 1. Python Best Practices (3.12+)
+- **Type Annotations:** Ensure completeness, modern syntax (`X | None`, builtin generics, `Self`, `type` statements), and avoid unparameterized containers or bare `Any`.
+- **Error Handling:** Identify bare/broad `except`, swallowed exceptions, missing `raise ... from`, and exceptions used for control flow.
+- **Resource Management:** Verify context managers for files, DB sessions, HTTP clients, and locks. Check for leaked tasks or connections.
+- **Data Modeling:** Check proper use of dataclasses vs. Pydantic models vs. dictionaries. Eliminate mutable default arguments and stringly-typed payloads.
+- **Idioms & Clean Code:** Verify `pathlib` usage over `os.path`, comprehensions vs manual loops, removal of dead code, and elimination of magic numbers.
+
+### 2. FastAPI
+- **Dependency Injection:** Verify `Depends` is used for shared resources (DB sessions, settings, clients) rather than global singletons.
+- **Route Design:** Validate HTTP verbs, status codes, path/query/body typing, `response_model`, and domain-based router organization.
+- **Lifecycle & Concurrency:** Ensure lifespan handlers are used instead of deprecated `@app.on_event`. Flag blocking synchronous calls in `async def` endpoints.
+
+### 3. NiceGUI
+- **Separation of Concerns:** Ensure UI components delegate business logic and persistence to service layers.
+- **Client State Handling:** Verify correct use of client-scoped state vs global state to avoid state leaks across sessions.
+- **Async Execution:** Check for blocking operations on the UI event loop and unbounded timers/pollers.
+
+### 4. Persistence (SQLModel / SQLAlchemy)
+- **Session Lifecycle:** Enforce one session per request/unit of work with explicit commit/rollback/close boundaries.
+- **Query Optimization:** Detect N+1 patterns, missing eager loads (`selectinload`/`joinedload`), queries inside loops, and unindexed filters.
+- **Cross-Dialect Portability:** Check compatibility for both SQLite (WAL mode, pragmas) and PostgreSQL (JSONB, locking, autoincrement).
+
+### 5. Pydantic V2 & Settings
+- **V2 Migration:** Flag legacy V1 patterns (`@validator`, `Config` class, `.dict()`, `parse_obj`) and use V2 equivalents (`@field_validator`, `model_config = ConfigDict(...)`, `model_dump()`).
+- **Settings Management:** Ensure `BaseSettings` is the single source of truth without scattered `os.getenv` calls or committed secrets.
+
+### 6. Concurrency & Asyncio Workers
+- **Task Lifecycle:** Flag unreferenced `create_task` calls that risk garbage collection, missing cancellation handling, and lack of graceful shutdown.
+- **Backpressure & Synchronization:** Check for appropriate use of `asyncio.Queue`, `TaskGroup`, `Lock`, and backoff retries.
+
+### 7. Provider Adapters (OpenRouter / APIs)
+- **Adapter Encapsulation:** Verify provider-specific details (headers, model names, payload formats) do not leak into UI or business logic.
+- **Client Lifecycle:** Reuse shared `AsyncClient` instances with proper connection pooling and timeouts. Validate API responses using Pydantic schemas.
+
+### 8. Testing & Quality Tooling
+- **Test Isolation:** Verify tests do not rely on live external services, real clocks, or shared global state.
+- **Async Test Setup:** Check `pytest-asyncio` configuration and fixture lifecycle.
+- **Project Test Contract (`pyproject.toml`):** `--strict-markers` is enabled, so every marker must be declared; `asyncio_mode = "strict"` requires explicit `@pytest.mark.asyncio`; declared markers are `unit`, `integration`, and `external`, and `external` must be excluded from default verification runs. `filterwarnings` promotes `coroutine ... was never awaited` to an **error** — treat any unawaited coroutine as a hard failure and a Critical/High finding, never a warning.
+- **Suite Signal Quality:** For low-value, redundant, or tautological tests, escalate to `.github/skills/test-effectiveness-auditor/skill.md` and fold its outcomes into the report.
+
+### 9. Duplication & Consolidation
+- Identify repeated code blocks, candidate helper abstractions, divergent patterns for identical operations, and duplicated domain constants.
+
+### 10. Orphaned/Dead Code Audit
+- Find candidate orphan modules/functions/classes with no inbound references.
+- Validate each candidate against dynamic wiring exceptions (entrypoints, plugin registration, reflection/dynamic imports, CLI hooks, test utilities).
+- Report outcomes as: removed orphan, retained-with-justification, or uncertain-follow-up.
+
+### 11. Architecture & Governance
+- **Architectural Drift:** Compare intended architecture rules against implementation behavior and cite concrete drift points.
+- **Systemic Health:** Evaluate domain cohesion, dependency direction, lifecycle consistency, and operational reliability seams.
+- **Invariant Routing:** Recommend the correct enforcement layer per rule (docs vs instructions vs skills vs tests).
+- **Meta-Tooling Alignment:** Recommend updates for instruction files and skills when repository patterns or contracts evolve.
+
+## Severity Rubric
+
+Severity reflects concrete consequence, never style preference or effort to fix.
+
+- **Critical:** Data or evidence loss/corruption; provenance or append-only history violated; secret leakage; silent wrong output presented as authoritative.
+- **High:** Architectural boundary violated (service/UI/persistence/provider); runtime failure or unhandled exception on a hot path (request handling, DB sessions, worker loop, external API calls); documented invariant contradicted by implementation.
+- **Medium:** Correctness risk under load or edge conditions (N+1, missing eager load, leaked task, missing timeout); drift between docs and code with no immediate runtime impact.
+- **Low:** Maintainability, typing completeness, duplication, naming, or dead code with no behavioral risk.
+
+### Reachability
+
+Severity states how bad the consequence is; **Reachability** states whether it can happen today.
+They are independent, and a finding is not complete without both. Record one of:
+
+- **Live:** reachable in the current configuration and deployment.
+- **Latent:** the defective code is present but unreachable because of a current setting, single-
+  instance deployment, or absent caller. **State the exact condition that unblocks it.**
+- **Theoretical:** requires a combination the project has explicitly ruled out.
+
+Latent findings carry a scheduling constraint that severity alone cannot express: a latent defect
+must usually be fixed *before* the change that makes it live, not after. Say so explicitly in the
+finding and reflect the ordering in the §9 action plan — for example, "fix the retry-category gate
+before raising `worker_max_retries` above 0," or "handle this `IntegrityError` before deploying a
+second worker replica." Do not downgrade severity merely because a finding is latent.
+
+### Conflicting invariants
+
+When a fix sits between two invariants that pull in opposite directions, say so in the
+**Recommendation** and name both, along with the test that guards each. Flag explicitly what the
+over-correction would be, because the simplest-looking fix usually satisfies one invariant by
+silently destroying the other. A recommendation that resolves one side without naming the other is
+incomplete and will be implemented incorrectly.
+
+## Output Report Structure & Template
+
+Generate Markdown reports at `./docs/reviews/<YYYY-MM-DD>-code-review.md` following this exact
+template structure. Reports are dated, non-canonical artifacts: `docs/reviews/**` is explicitly
+**not** part of the canonical authority set that the canonical-authority check resolves against.
+
+```markdown
+# Architecture & Code Review Report
+
+**Repository Target:** `project-root/`
+**Target Stack:** Python 3.12+ | FastAPI | NiceGUI | SQLModel/SQLAlchemy | Pydantic V2 | asyncio | OpenRouter
+
+---
+
+## 1. Executive Summary
+- 5-10 bullets on overall health, top risks, and high-leverage refactors.
+
+---
+
+## 2. Executive Architecture Assessment
+- High-level verdict on domain cohesion, boundary clarity, and architecture fitness.
+- Top 3-5 systemic risks or bottlenecks.
+
+---
+
+## 3. Findings by Severity
+
+### Critical Severity
+#### [CRIT-01] Title
+- **Location:** `path/to/file.py:lines`
+- **Reachability:** Live / Latent (state the exact condition that unblocks it) / Theoretical
+- **Problem & Consequence:** Concrete consequence, not a style opinion.
+- **Blast Radius:** Every consumer of the symbols the recommendation changes, each confirmed
+  safe. Write `None — change is local` only after actually searching. If the fix touches a
+  shared field or helper, list the call sites (including tests and evidence/logging paths).
+- **Recommendation:** Fix with before/after sketch. If two invariants conflict here, name both,
+  name the test guarding each, and state what the over-correction would be.
+- **Effort:** S / M / L
+
+### High Severity
+#### [HIGH-01] Title
+...
+
+### Medium Severity
+#### [MED-01] Title
+...
+
+### Low Severity
+#### [LOW-01] Title
+...
+
+---
+
+## 4. Architectural Drift & Gap Analysis
+
+`Direction` is `doc->code` (implementation must change to match documented intent) or
+`code->doc` (an undocumented but repeatable convention that should be formalized).
+
+| Area / Component | Direction | Documented / Intended Rule | Actual Implementation State | Severity | Recommended Resolution |
+| :--- | :--- | :--- | :--- | :--- | :--- |
+
+---
+
+## 5. Invariant Inventory & Routing Recommendations
+| Invariant / Constraint | Current Location | Recommended Target Layer | Rationale |
+| :--- | :--- | :--- | :--- |
+
+---
+
+## 6. Stack-Specific Analysis
+- Python 3.12+ Best Practices
+- FastAPI
+- NiceGUI
+- SQLModel & SQLAlchemy
+- Pydantic V2 & Settings
+- Asyncio Workers
+- OpenRouter / Adapter Boundary
+- Testing & Quality Tooling
+
+---
+
+## 7. Duplication & Consolidation Report
+| Pattern / Duplication | Locations | Proposed Canonical Home | Estimated Lines Removed |
+| :--- | :--- | :--- | :--- |
+
+### Proposed Canonical Abstractions
+- Code signatures and implementation homes.
+
+---
+
+## 8. Meta-Tooling & Instruction Update Recommendations
+- Required updates to docs, instructions, skills, or tests to keep enforcement current.
+
+---
+
+## 9. Prioritized Dependency-Ordered Action Plan
+1. **Phase 1: Blocking fixes**
+2. **Phase 2: Enforcement hardening**
+3. **Phase 3: Reliability & concurrency**
+4. **Phase 4: Consolidation & refactoring**
+5. **Phase 5: Non-blocking governance/documentation depth**
+
+---
+
+## 10. Preserved Strengths
+- Existing patterns worth maintaining.

.github/skills/test-effectiveness-auditor/skill.md

@@ -0,0 +1,98 @@
+---
+name: test-effectiveness-auditor
+description: Periodic reviewer for test-suite signal quality. Detects low-value or redundant tests, validates contract coverage, and recommends pruning or strengthening actions.
+---
+
+# Test Effectiveness Auditor
+
+Run a deterministic audit of test usefulness. Focus on whether tests catch real regressions, not whether they merely execute code.
+
+## When to Use
+
+- Monthly/quarterly test-health review.
+- Pre-release hardening when test count grows quickly.
+- After major AI-assisted test generation.
+- When suite runtime is increasing without clear quality gains.
+
+## Primary Objectives
+
+1. Identify tests that are weak, redundant, or non-diagnostic.
+2. Confirm critical contracts are guarded by meaningful assertions.
+3. Produce a prune/strengthen backlog with explicit risk and effort.
+
+## Normative References (Transcription Repo)
+
+1. `docs/*`
+2. `docs/invariant/*`
+3. `.github/instructions/*.instructions.md`
+4. `tests/test_meta_contract_guards.py`
+5. Contract-specific guards (`tests/test_service_boundaries.py`, `tests/test_ui_boundaries.py`, worker/evidence/media/error suites)
+
+## Deterministic Audit Checks
+
+### A. Contract Traceability
+- Each high-risk contract maps to at least one focused regression test file.
+- Missing mapping is a gap.
+
+### B. Assertion Strength
+- Flag tests that only assert status code, non-null, or “no exception” without validating state transitions or persisted outcomes.
+- Prefer assertions on domain effects: DB rows, status changes, error categories, evidence writes, or emitted payload shape.
+
+### C. Failure-Path Coverage
+- Critical paths must include negative-path tests (timeouts, provider errors, validation failures, cancellation paths, retries).
+- Happy-path-only coverage on critical modules is a gap.
+
+### D. Redundancy and Noise
+- Detect near-duplicate tests asserting the same behavior at multiple layers with no extra signal.
+- Recommend canonical location (unit/integration) and prune overlaps.
+
+### E. Mutation/Change Sensitivity
+- Prefer mutation testing for high-risk modules when practical.
+- If not run, identify tests likely to survive meaningful code mutations (low sensitivity).
+
+### F. Drift Guards
+- Verify config/doc/instruction contracts have deterministic guards and are current.
+- Ensure settings/docs synchronization checks remain active.
+
+## Evidence Standards
+
+- Every finding must include concrete file paths and line ranges.
+- No speculative claims.
+- Distinguish clearly between:
+  - **Confirmed ineffective tests**
+  - **Likely weak tests (needs mutation/probe confirmation)**
+
+## Output Format
+
+Produce a Markdown report at `docs/reviews/<YYYY-MM-DD>-test-effectiveness.md`, using today's date.
+Like code review reports, it is a dated, non-canonical artifact: `docs/reviews/**` is not part of the
+canonical authority set.
+
+```markdown
+# Test Effectiveness Audit Report
+
+## 1. Executive Verdict
+- Effective / Effective with Conditions / Needs Remediation
+- Top risks to confidence
+
+## 2. Contract Coverage Matrix
+| Contract | Guarding Tests | Signal Quality | Gap | Action |
+| :--- | :--- | :--- | :--- | :--- |
+
+## 3. Weak/Redundant Test Findings
+| Finding ID | Location | Why Low-Signal | Risk | Recommendation |
+| :--- | :--- | :--- | :--- | :--- |
+
+## 4. Prune/Strengthen Backlog
+| Task ID | Goal | Files | Acceptance Criteria | Validation |
+| :--- | :--- | :--- | :--- | :--- |
+
+## 5. Confidence Recommendation
+- Go / Go with Conditions / No-Go for release confidence
+```
+
+## Decision Rules
+
+- Do not recommend deleting a test unless equivalent or stronger coverage is identified.
+- Prefer strengthening assertions before adding more tests.
+- Prioritize deterministic contract guards over broad snapshot-style tests.

.github/workflows/quality-gate.yml

@@ -0,0 +1,50 @@
+name: Quality Gate
+
+# Repository quality gate. Before this workflow existed, ruff, ty, and pytest were
+# enforced only by .pre-commit-config.yaml for developers who had run
+# `pre-commit install`.
+
+on:
+  push:
+  pull_request:
+
+jobs:
+  gate:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out the commit under test
+        uses: actions/checkout@v4
+
+      - name: Install uv
+        run: |
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          echo "$HOME/.local/bin" >> "$GITHUB_PATH"
+
+      - name: Install dependencies from the lockfile
+        # --locked fails if uv.lock has drifted from pyproject.toml, so a stale
+        # lockfile is caught here rather than producing an untested dependency set.
+        run: uv sync --locked
+
+      - name: Write placeholder configuration
+        # Settings requires openrouter_api_key and 115 tests cannot construct
+        # Settings without it. This is written to .env.production rather than exported
+        # as an environment variable on purpose: the external tests guard on
+        # os.getenv("OPENROUTER_API_KEY"), which reads the process environment and
+        # not the file, so writing the file reproduces the local result exactly -
+        # the 4 external tests skip instead of running against a fake key and
+        # failing. Exporting it instead produces 3 failures.
+        run: echo "OPENROUTER_API_KEY=ci-placeholder-not-a-real-key" > .env.production
+
+      - name: Lint and type check
+        # Runs the hooks defined in .pre-commit-config.yaml instead of repeating
+        # "ruff check" and "ty check" here. The commands then have one definition,
+        # so the local and CI gates cannot drift apart.
+        run: uv run pre-commit run --all-files --show-diff-on-failure
+
+      - name: Tests
+        # Deliberately unfiltered, unlike the "-m 'not external'" form the guidance files
+        # use for local runs. Tests marked "external" skip themselves when live-service
+        # credentials are absent, so CI gets the same effective set plus a real run of any
+        # external test whose credentials are configured. Not drift -- do not "fix" this to
+        # match the local command without also giving those tests a way to run.
+        run: uv run pytest

.gitignore

@@ -11,9 +11,25 @@ wheels/
 
 # Environment secrets
 .env
+.env.production
 
 # SQLite database
 *.db
 
-# Document images
-uploads/*
+# All data including db, backups, document images, photos, and logs:
+data/*
+data/backups/*
+data/documents/*
+data/logs/*
+data/photos/*
+data-local/*
+
+
+# Migration tests
+data-migration-test/*
+.migration-bundle/*
+
+# Cloudflare tunnel local runtime files
+deploy/cloudflared/config.yml
+deploy/cloudflared/config.yaml
+deploy/cloudflared/credentials.json

.pre-commit-config.yaml

@@ -0,0 +1,29 @@
+# Quality gate: `ruff check`, `ruff format --check`, and `ty check`
+# are blocking once known `ty` false positives are suppressed inline with rationale.
+#
+# Both tools are uv-managed dev dependencies and are not on PATH, so each entry must
+# go through `uv run`.
+repos:
+  - repo: local
+    hooks:
+      - id: ruff
+        name: ruff check
+        entry: uv run ruff check
+        language: system
+        types_or: [python, pyi]
+        require_serial: true
+      - id: ruff-format
+        name: ruff format check
+        entry: uv run ruff format --check .
+        language: system
+        types_or: [python, pyi]
+        pass_filenames: false
+        require_serial: true
+      - id: ty
+        name: ty check
+        entry: uv run ty check
+        language: system
+        types_or: [python, pyi]
+        pass_filenames: false
+        require_serial: true
+        verbose: true

.vscode/launch.json

@@ -8,14 +8,10 @@
             "module": "debugpy",
             "args": [
                 "-m",
-                "uvicorn",
-                "transcription.app:create_app",
-                "--factory",
-                "--host",
-                // "127.0.0.1",
-                "0.0.0.0",
-                "--port",
-                "8080"
+                "transcription",
+                "--host", "127.0.0.1",
+                "--port", "9999",
+                "--database.driver", "sqlite"
             ],
             "justMyCode": true,
             "console": "integratedTerminal",

.vscode/settings.json

@@ -0,0 +1,3 @@
+{
+    "chat.sessionSync.enabled": true
+}

AGENTS.md

@@ -0,0 +1,117 @@
+# AGENTS.md
+
+Orientation for AI agents working in this repository. This file is a **router**, not a spec:
+it points at canonical authority and flags the traps that are expensive to discover by trial.
+Where this file and `docs/*` disagree, `docs/*` wins.
+
+## What This Is
+
+A document transcription system that preserves durable archival records (Documents, Sources,
+People) and executes page transcription asynchronously through vision/LLM providers. Its
+defining constraint is **evidence**: every machine attempt is recorded append-only with
+request/response provenance. Features that would lose, mutate, or obscure that history are
+wrong regardless of how convenient they are.
+
+Stack: Python 3.12+ · FastAPI + NiceGUI · SQLModel/SQLAlchemy (SQLite-first, PostgreSQL-
+compatible) · Pydantic V2 · asyncio worker · OpenRouter adapter.
+
+## Commands
+
+This is a `uv` project. **Nothing is on `PATH`** — `ruff`, `ty`, and `pytest` all require
+`uv run`. Bare invocations fail with command-not-found.
+
+```bash
+uv run ruff check .                  # lint (blocking in pre-commit)
+uv run ruff format --check .         # format (blocking in pre-commit)
+uv run ty check                      # types (blocking in pre-commit)
+uv run pytest -q -m "not external"   # default verification run
+```
+
+`external` marks tests that hit live services; always exclude it unless explicitly asked.
+All four commands are expected to pass clean — there is no tolerated baseline of failures.
+If `ty` reports something, fix it or suppress it inline *with a rationale comment*; a bare
+`ignore` will not survive review.
+
+## Authority Order
+
+Resolve every question in this order, and stop at the first that answers it:
+
+1. **`docs/*`** — canonical. Start at [`docs/index.md`](docs/index.md), which defines the
+   reading order. `docs/invariant/*` holds cross-version rules that outlive any release.
+2. **`.github/instructions/*.md`** — active steering, auto-attached when you edit matching
+   paths. Covers services, UI, providers, tests, error handling, and documentation sync.
+3. **`.github/skills/*`** — periodic audit procedures (code review, provenance, test
+   effectiveness).
+4. **`tests/`** — deterministic enforcement. A guard test is the ground truth for whatever
+   rule it encodes.
+
+`.github/agents/` and `.github/prompts/` hold named workflows that are loaded only when
+invoked explicitly, so they never override the order above. They are how a review or audit
+is *started*, not a source of rules.
+
+`docs/reviews/**` is **not** canonical. Those are dated, opinionated snapshots that were
+accurate when written and may since have been fixed, superseded, or found wrong.
+
+## Layout
+
+| Path | Role |
+| :--- | :--- |
+| `src/transcription/ui/**`, `api/**` | Interface. No direct persistence access. |
+| `src/transcription/services/**` | Domain logic and transaction ownership. |
+| `src/transcription/db/**` | Models and persistence. |
+| `src/transcription/providers/**` | Provider adapters; provider details stop here. |
+| `src/transcription/worker.py` | Asyncio worker loop. |
+| `tests/` | Includes boundary/contract guards, not just behavior tests. |
+
+## Enforced Boundaries
+
+These are not conventions — a test fails if you break them:
+
+- **No service-to-service imports** (`test_service_boundaries.py`). Compose in the caller.
+- **No persistence access from pages/components** (`test_ui_boundaries.py`, allowlist-based).
+- **No hand-rolled error notifications in UI** — use the shared error presenter.
+- **No stringly-typed status literals** — use the enums (`test_model_contract_guards.py`).
+- **Attempt history is append-only** (`test_evidence_provenance.py`).
+- **`docs/schema.md` stays field-accurate** with `db/models.py`.
+- **Orphans are tracked, not tolerated** — `test_orphan_sweep.py` records each retained
+  orphan with rationale in `KNOWN_ORPHANS`.
+
+## Traps
+
+Non-obvious things that have already caused real bugs here:
+
+- **`AppError.message` vs `AppError.detail`.** `message` is user/API-facing and must stay
+  generic — never put exception text or filesystem paths in it. `detail` is internal-only and
+  is what reaches logs and `ExecutionAttempt.error_detail`. Putting root-cause data in
+  `message` leaks; removing it from `detail` silently degrades provenance. See
+  `docs/error_handling.md`.
+- **Two competing atomicity invariants in `services/workflows.py`.** Intermediate pages must
+  commit individually (durability across a long multi-page job); the *final* page must commit
+  atomically with the terminal job status. Collapsing the batch into one transaction satisfies
+  the second and destroys the first. Both are guarded — `test_workflows_reliability.py` and
+  `tests/integration/test_pipeline_atomicity.py`.
+- **Shared symbols have more consumers than the obvious one.** Before changing a model field,
+  exception attribute, or helper return value, grep for every consumer including tests.
+  Evidence and logging paths frequently read the same fields the UI does.
+- **`Tag` is owned by two services, on purpose.** Every other model has exactly one owning
+  service, so the ownership rule reads as absolute — it isn't. `Tag` is a single table reached
+  through two `RegistryService[Tag]` facades, `TagRegistry` (documents) and `PersonTagRegistry`
+  (people), which count usage through `DocumentTag` and `PersonTag` respectively. Changing tag
+  semantics through one facade silently changes the other. Consolidating them under one service
+  is not a cleanup; it makes the other side a cross-aggregate writer.
+- **Import style:** ruff `isort` runs with `force-single-line = true`. One import per line.
+- **Latent defects have ordering constraints.** Some code is unreachable only because of a
+  current setting or single-instance deployment. Fix it *before* the change that unblocks it,
+  not after.
+
+## Change Protocol
+
+- **Write the failing test first** for behavioral fixes, and confirm it actually fails for the
+  reason you think. Several bugs here were subtle enough that a test written afterward would
+  have passed against the broken code.
+- **Update docs in the same change** when you alter a contract, behavior, or scope — see
+  `.github/instructions/documentation-sync.instructions.md`.
+- **Do not commit unless asked.** Making a requested change is not consent to commit it.
+- **Do not push or open PRs on your own initiative.**
+- **Scope discipline:** fix what was asked plus what your change genuinely breaks. Pre-existing
+  unrelated issues are a separate conversation.

Dockerfile

@@ -13,6 +13,8 @@ RUN uv sync --frozen --no-dev --no-install-project
 
 COPY src ./src
 COPY prompts ./prompts
+COPY tools ./tools
+COPY deploy ./deploy
 RUN uv sync --frozen --no-dev
 
 
@@ -27,12 +29,18 @@ ENV PYTHONDONTWRITEBYTECODE=1 \
 
 WORKDIR /app
 
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends postgresql-client \
+    && rm -rf /var/lib/apt/lists/*
+
 RUN groupadd --system --gid 1001 appgroup \
     && useradd --system --uid 1001 --gid appgroup --create-home appuser
 
 COPY --from=builder /app/.venv /app/.venv
 COPY --from=builder /app/src /app/src
 COPY --from=builder /app/prompts /app/prompts
+COPY --from=builder /app/tools /app/tools
+COPY --from=builder /app/deploy /app/deploy
 
 RUN mkdir -p /app/uploads /app/data \
     && chown -R appuser:appgroup /app

README.md

@@ -22,18 +22,84 @@ uv sync
 
 ### 2) Configure environment
 
-Create a `.env` file in the project root (minimum required setting shown):
+Create a `.env.production` file in the project root with the required OpenRouter API key:
 
 ```env
 OPENROUTER_API_KEY=your_openrouter_api_key
 ```
 
-Optional settings (defaults shown):
+Settings are read from CLI arguments first, then environment variables, then `.env.production`, then the defaults below.
+
+### Configuration Source Precedence
+
+When the same setting is provided in multiple places, the value is chosen in this order (highest priority first):
+
+1. CLI arguments (for example `--port 8000`)
+2. Settings constructor arguments (used mainly in tests)
+3. Environment variables
+4. `.env.production` file values
+5. Model defaults in code
+
+Practical examples:
+
+- `--port 8000` overrides both `PORT=8000` in the shell and `PORT=7000` in `.env.production`.
+- `DATABASE__PATH=prod.db` in the shell overrides `DATABASE__PATH=dev.db` in `.env.production`.
+
+#### Server and runtime
+
+| Environment variable | Default | Description |
+| --- | --- | --- |
+| `HOST` | `0.0.0.0` | Address on which the server listens. |
+| `PORT` | `8000` | Server port. |
+| `LOG_LEVEL` | `info` | Uvicorn and application log level. |
+| `RELOAD` | `false` | Restart the development server when source files change. |
+| `ENVIRONMENT` | `development` | Runtime environment: `development`, `test`, or `production`. |
+| `RUN_EMBEDDED_WORKER` | `true` | Run worker loop inside web app process. Set `false` when using a dedicated worker service. |
+
+#### Provider
+
+| Environment variable | Default | Description |
+| --- | --- | --- |
+| `PROVIDER` | `openrouter` | Transcription provider. |
+| `OPENROUTER_API_KEY` | Required | OpenRouter API key. |
+| `PROVIDER_MODEL` | Provider default | Optional model override. |
+| `OPENROUTER_HTTP_REFERER` | Unset | Optional OpenRouter attribution URL. |
+| `OPENROUTER_APP_TITLE` | Unset | Optional OpenRouter attribution title. |
+
+#### Database and files
+
+Use nested env vars for database settings (recommended):
 
 ```env
-DATABASE_URL=sqlite:///./transcription.db
+DATABASE__DRIVER=sqlite
+DATABASE__PATH=app.db
+# BOOTSTRAP_SCHEMA_ON_STARTUP=true
+SQLITE_CHECK_SAME_THREAD=false
 UPLOAD_DIR=./uploads
 PROMPT_DIR=./prompts
+DEFAULT_PROMPT_NAME=transcribe_document.md
+# TRANSCRIPTION_TEMPERATURE=0.2  # range: 0.0-2.0
+# TRANSCRIPTION_TOP_P=0.9       # range: 0.0-1.0
+```
+
+For PostgreSQL:
+
+```env
+DATABASE__DRIVER=postgres
+DATABASE__HOST=localhost
+DATABASE__PORT=5432
+DATABASE__DATABASE=transcription
+DATABASE__USER=postgres
+DATABASE__PASSWORD=change-me
+```
+
+This uses Pydantic nested settings (`env_nested_delimiter='__'`) and avoids JSON blobs in env files. A top-level `DATABASE={...}` JSON value is still supported as a fallback, and nested keys such as `DATABASE__PATH` take precedence over conflicting JSON keys.
+
+`BOOTSTRAP_SCHEMA_ON_STARTUP` creates missing tables when the app starts. When unset, it is enabled in `development` and `test`, and disabled in `production`; set it explicitly to override that policy. `SQLITE_CHECK_SAME_THREAD` defaults to `false`.
+
+#### Worker
+
+```env
 WORKER_MAX_RETRIES=0
 WORKER_RETRY_BACKOFF_SECONDS=0
 WORKER_PROVIDER_TIMEOUT_SECONDS=20
@@ -45,13 +111,44 @@ WORKER_FAIL_ON_FINISH_REASON_LENGTH=false
 ### 3) Run the app
 
 ```bash
-uv run uvicorn transcription.app:create_app --factory --reload
+uv run python -m transcription --port 8000 --reload --database.driver sqlite --bootstrap-schema-on-startup
 ```
 
+This starts the development server with SQLite, creates missing tables, and enables automatic reload. Run `uv run python -m transcription --help` for all CLI options; CLI names use kebab case and nested database options use dot notation, such as `--database.path ./data/transcription.db`.
+
 ### 4) Open in browser
 
-- GUI: [http://[IP_ADDRESS]:8000/ui](http://[IP_ADDRESS]:8000/ui)
-- Health check: [http://[IP_ADDRESS]:8000/healthz](http://[IP_ADDRESS]:8000/healthz)
+- GUI: [http://localhost:8000/ui](http://localhost:8000/ui)
+- Health check: [http://localhost:8000/healthz](http://localhost:8000/healthz)
+
+Replace `localhost` with the server's hostname or IP address when connecting from another machine.
+
+## Production stack (Phase 1)
+
+Use the production compose profile for split app/worker deployment with PostgreSQL and Cloudflare Tunnel:
+
+```bash
+copy .env.production.example .env.production
+docker compose -f docker-compose.production.yml up -d --build
+```
+
+Services:
+
+- `app`: FastAPI + NiceGUI runtime (`RUN_EMBEDDED_WORKER=false`)
+- `worker`: standalone queue processor (`python -m transcription.worker_service`)
+- `postgres`: primary datastore
+- `cloudflared`: tunnel client using mounted ingress config + `CLOUDFLARE_TUNNEL_TOKEN`
+
+Operational defaults in the production compose file:
+
+- worker healthcheck is disabled (the worker process has no HTTP `/healthz` endpoint)
+- cloudflared is pinned to HTTP/2 with explicit DNS resolvers (`1.1.1.1`, `1.0.0.1`) for restricted LXC/container DNS environments
+
+Cloudflare setup files:
+
+1. `copy deploy\cloudflared\config.yml.example deploy\cloudflared\config.yml`
+2. set `CLOUDFLARE_TUNNEL_TOKEN` in `.env.production`
+3. update ingress hostnames in `deploy\cloudflared\config.yml`
 
 ## How to navigate the GUI
 
@@ -72,7 +169,74 @@ uv run uvicorn transcription.app:create_app --factory --reload
 
 ## Prompt artifacts
 
-Prompt files are stored in `prompts/` and loaded from `PROMPT_DIR` (default: `./prompts`).
+Prompt files are stored directly in `PROMPT_DIR` (default: `./prompts`). `DEFAULT_PROMPT_NAME` must be a filename,
+not a path. Each job snapshots the validated prompt text, SHA-256 hash, and sampling values for reproducibility.
 
 The canonical MVP prompt is:
 - `prompts/transcribe_document.md`
+
+## Database migration workflow
+
+Schema upgrades use an explicit export/import rebuild flow (no runtime legacy write compatibility).  
+See `docs/data_migration.md` for commands and cutover steps.
+
+## Backup and restore workflow
+
+Production backup/restore (PostgreSQL + uploads + deployment config) steps are documented in `docs/backup_restore.md`.
+
+## Destructive test procedure (with data backup)
+
+AI execution policy: before the first unit-test run in a test/fix cycle, create one backup of `./data`. Reuse that same backup for every subsequent test run in the cycle. After tests succeed, always pause and ask whether to restore now.
+
+Use the cross-platform Python wrapper below whenever an AI agent runs tests against this repository.
+
+1. Create one backup of `./data` and mark it as the active test-cycle backup.
+2. Run your test command.
+3. On failure, fix the errors and run the wrapper again; it reuses the active backup and never backs up post-test data.
+4. On success, always prompt whether to restore now (do not auto-restore unless explicitly approved).
+5. Close the cycle only by restoring the active backup or explicitly accepting the current data.
+
+Preflight behavior:
+
+- Backup preflight is warning-only when `data/transcription.db` appears in use.
+- Restore preflight is blocking: the script prompts you to close conflicting applications, then type `retry` to re-check or `cancel` to skip restore.
+
+### Run with confirmation-gated restore (default)
+
+```bash
+uv run python tools/run_destructive_tests.py -- pytest tests/services/test_job_service.py tests/ui/test_jobs_page.py
+```
+
+After tests pass, the script asks whether to restore backup immediately.
+
+This is the required default mode for AI-assisted test runs because it gives time to verify and accept code changes before any restoration happens.
+
+### Run with automatic restore (non-interactive)
+
+```bash
+uv run python tools/run_destructive_tests.py --auto-restore -- pytest
+```
+
+### Run without terminal prompt (decide restore later)
+
+```bash
+uv run python tools/run_destructive_tests.py --skip-restore-prompt -- pytest
+```
+
+This keeps both the current post-test state and the backup, so restore can be decided explicitly later.
+
+Repeated wrapper invocations reuse the backup recorded in `.test-backups/.active-backup`. If that backup is missing, the wrapper stops rather than creating a replacement from potentially destructive post-test data.
+
+### Restore later from a saved backup
+
+```bash
+uv run python tools/run_destructive_tests.py --restore-from data-backup-YYYYMMDD-HHMMSS
+```
+
+To keep the current data and close the active cycle without restoring:
+
+```bash
+uv run python tools/run_destructive_tests.py --accept-current-data
+```
+
+Backups are stored in `.test-backups/` and ignored by git.

deploy/backup/create_postgres_backup.sh

@@ -0,0 +1,69 @@
+#!/usr/bin/env sh
+set -eu
+
+BACKUP_DIR="${BACKUP_DIR:-./data/backups}"
+RETENTION_DAYS="${BACKUP_RETENTION_DAYS:-14}"
+UPLOAD_DIR="${UPLOAD_DIR:-/app/uploads}"
+PROMPT_DIR="${PROMPT_DIR:-/app/prompts}"
+DATABASE_DRIVER="${DATABASE__DRIVER:-postgres}"
+DATABASE_HOST="${DATABASE__HOST:-postgres}"
+DATABASE_PORT="${DATABASE__PORT:-5432}"
+DATABASE_NAME="${DATABASE__DATABASE:-}"
+DATABASE_USER="${DATABASE__USER:-}"
+DATABASE_PASSWORD="${DATABASE__PASSWORD:-}"
+
+timestamp="$(date -u +%Y%m%d-%H%M%S)"
+postgres_file="postgres-${timestamp}.dump"
+manifest_file="backup-${timestamp}.manifest"
+
+mkdir -p "${BACKUP_DIR}"
+if [ "${DATABASE_DRIVER}" != "postgres" ]; then
+  echo "create_postgres_backup.sh requires DATABASE__DRIVER=postgres." >&2
+  exit 1
+fi
+if [ -z "${DATABASE_NAME}" ] || [ -z "${DATABASE_USER}" ] || [ -z "${DATABASE_PASSWORD}" ]; then
+  echo "DATABASE__DATABASE, DATABASE__USER, and DATABASE__PASSWORD must be set." >&2
+  exit 1
+fi
+if ! command -v pg_dump >/dev/null 2>&1; then
+  echo "pg_dump is not installed in this environment." >&2
+  exit 1
+fi
+PGPASSWORD="${DATABASE_PASSWORD}" pg_dump \
+  -h "${DATABASE_HOST}" \
+  -p "${DATABASE_PORT}" \
+  -U "${DATABASE_USER}" \
+  -d "${DATABASE_NAME}" \
+  -Fc \
+  > "${BACKUP_DIR}/${postgres_file}"
+
+cat > "${BACKUP_DIR}/${manifest_file}" <<EOF
+created_at_utc=${timestamp}
+postgres_dump=${postgres_file}
+backup_dir=${BACKUP_DIR}
+uploads_backup_dir=${BACKUP_DIR}/uploads
+prompts_backup_dir=${BACKUP_DIR}/prompts
+EOF
+
+find "${BACKUP_DIR}" -type f \( \
+  -name 'postgres-*.dump' -o \
+  -name 'backup-*.manifest' \
+\) -mtime +"${RETENTION_DAYS}" -delete
+
+uploads_backup_dir="${BACKUP_DIR}/uploads"
+mkdir -p "${uploads_backup_dir}"
+if [ -d "${UPLOAD_DIR}" ]; then
+  cp -an "${UPLOAD_DIR}/." "${uploads_backup_dir}/"
+fi
+
+prompts_backup_dir="${BACKUP_DIR}/prompts"
+mkdir -p "${prompts_backup_dir}"
+if [ -d "${PROMPT_DIR}" ]; then
+  cp -a "${PROMPT_DIR}/." "${prompts_backup_dir}/"
+fi
+
+echo "Created backup set:"
+echo "  ${BACKUP_DIR}/${postgres_file}"
+echo "  ${BACKUP_DIR}/${manifest_file}"
+echo "  ${uploads_backup_dir}/ (incremental uploads mirror)"
+echo "  ${prompts_backup_dir}/ (prompts mirror)"

deploy/backup/mount_synology_cifs.example.sh

@@ -0,0 +1,36 @@
+#!/usr/bin/env sh
+set -eu
+
+# Example only. Copy to a local script and replace placeholder values.
+# Do NOT commit secrets.
+
+SHARE="//nas-host-or-ip/share-name"
+MOUNT_POINT="/mnt/nas-backups"
+CREDENTIALS_FILE="/etc/samba/credentials/nas-share-credentials"
+USERNAME="replace-with-nas-user"
+PASSWORD="replace-with-nas-password"
+
+if [ "${USERNAME}" = "replace-with-nas-user" ] || [ "${PASSWORD}" = "replace-with-nas-password" ]; then
+  echo "Edit USERNAME and PASSWORD placeholders before running this script."
+  exit 1
+fi
+
+mkdir -p "${MOUNT_POINT}"
+mkdir -p "$(dirname "${CREDENTIALS_FILE}")"
+
+cat > "${CREDENTIALS_FILE}" <<'EOF'
+username=__USERNAME__
+password=__PASSWORD__
+EOF
+sed -i "s|__USERNAME__|${USERNAME}|g" "${CREDENTIALS_FILE}"
+sed -i "s|__PASSWORD__|${PASSWORD}|g" "${CREDENTIALS_FILE}"
+chmod 600 "${CREDENTIALS_FILE}"
+
+mount -t cifs "${SHARE}" "${MOUNT_POINT}" \
+  -o "credentials=${CREDENTIALS_FILE},vers=3.0,iocharset=utf8,uid=0,gid=0,file_mode=0600,dir_mode=0700"
+
+echo ""
+echo "Mounted ${SHARE} at ${MOUNT_POINT}"
+echo ""
+echo "To persist across reboot, add this line to /etc/fstab:"
+echo "${SHARE} ${MOUNT_POINT} cifs credentials=${CREDENTIALS_FILE},vers=3.0,iocharset=utf8,uid=0,gid=0,file_mode=0600,dir_mode=0700,_netdev,nofail,x-systemd.automount 0 0"

deploy/backup/restore_postgres_backup.sh

@@ -0,0 +1,86 @@
+#!/usr/bin/env sh
+set -eu
+
+if [ "$#" -lt 1 ]; then
+  echo "Usage: $0 <path-to-postgres-dump>"
+  exit 1
+fi
+
+dump_file="$1"
+COMPOSE_FILE="${COMPOSE_FILE:-docker-compose.production.yml}"
+ENV_FILE="${ENV_FILE:-.env.production}"
+SYNOLOGY_BACKUP_DIR="${SYNOLOGY_BACKUP_DIR:-}"
+
+if [ ! -f "${dump_file}" ]; then
+  echo "Backup file not found: ${dump_file}"
+  exit 1
+fi
+
+backup_dir="$(dirname "${dump_file}")"
+backup_name="$(basename "${dump_file}")"
+timestamp="$(printf '%s' "${backup_name}" | sed -n 's/^postgres-\([0-9]\{8\}-[0-9]\{6\}\)\.dump$/\1/p')"
+uploads_file=""
+config_file=""
+
+# If SYNOLOGY_BACKUP_DIR wasn't exported in the shell, read it from ENV_FILE.
+if [ -z "${SYNOLOGY_BACKUP_DIR}" ] && [ -f "${ENV_FILE}" ]; then
+  SYNOLOGY_BACKUP_DIR="$(
+    sed -n 's/^SYNOLOGY_BACKUP_DIR=//p' "${ENV_FILE}" | tail -n 1
+  )"
+fi
+
+if [ -n "${timestamp}" ]; then
+  # Legacy local full-archive naming.
+  candidate_uploads="${backup_dir}/uploads-${timestamp}.tar.gz"
+  candidate_config="${backup_dir}/config-${timestamp}.tar.gz"
+  if [ -f "${candidate_uploads}" ]; then
+    uploads_file="${candidate_uploads}"
+  fi
+  if [ -f "${candidate_config}" ]; then
+    config_file="${candidate_config}"
+  fi
+fi
+
+docker compose --env-file "${ENV_FILE}" -f "${COMPOSE_FILE}" exec -T postgres sh -lc \
+  "PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -U \"\$POSTGRES_USER\" -d postgres -c \"DROP DATABASE IF EXISTS \\\"\$POSTGRES_DB\\\";\""
+
+docker compose --env-file "${ENV_FILE}" -f "${COMPOSE_FILE}" exec -T postgres sh -lc \
+  "PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -U \"\$POSTGRES_USER\" -d postgres -c \"CREATE DATABASE \\\"\$POSTGRES_DB\\\";\""
+
+cat "${dump_file}" | docker compose --env-file "${ENV_FILE}" -f "${COMPOSE_FILE}" exec -T postgres sh -lc \
+  "PGPASSWORD=\"\$POSTGRES_PASSWORD\" pg_restore -U \"\$POSTGRES_USER\" -d \"\$POSTGRES_DB\" --clean --if-exists --no-owner --no-privileges"
+
+if [ -n "${SYNOLOGY_BACKUP_DIR}" ] && [ -d "${SYNOLOGY_BACKUP_DIR}/uploads" ]; then
+  docker compose --env-file "${ENV_FILE}" -f "${COMPOSE_FILE}" run --rm --no-deps \
+    -v "${SYNOLOGY_BACKUP_DIR}:/backup" \
+    --entrypoint sh app -lc \
+    "mkdir -p /app/uploads/documents /app/uploads/photos && \
+    find /app/uploads/documents -mindepth 1 -delete && \
+    find /app/uploads/photos -mindepth 1 -delete && \
+    if [ -d /backup/uploads/documents ]; then cp -a /backup/uploads/documents/. /app/uploads/documents/; fi && \
+    if [ -d /backup/uploads/photos ]; then cp -a /backup/uploads/photos/. /app/uploads/photos/; fi && \
+    if [ -f /backup/uploads/homepage.md ]; then cp /backup/uploads/homepage.md /app/uploads/homepage.md; else rm -f /app/uploads/homepage.md; fi"
+elif [ -n "${uploads_file}" ]; then
+  # Legacy local full-archive restore.
+  cat "${uploads_file}" | docker compose --env-file "${ENV_FILE}" -f "${COMPOSE_FILE}" run --rm --no-deps --entrypoint sh app -lc \
+    "mkdir -p /app/uploads && find /app/uploads -mindepth 1 -delete && tar -xzf - -C /app/uploads"
+fi
+
+if [ -n "${SYNOLOGY_BACKUP_DIR}" ] && [ -n "${timestamp}" ] && [ -f "${SYNOLOGY_BACKUP_DIR}/config-${timestamp}.tar.gz" ]; then
+  tar -xzf "${SYNOLOGY_BACKUP_DIR}/config-${timestamp}.tar.gz" -C .
+elif [ -n "${config_file}" ]; then
+  # Legacy local full-archive restore.
+  tar -xzf "${config_file}" -C .
+fi
+
+echo "Restore complete from: ${dump_file}"
+if [ -n "${SYNOLOGY_BACKUP_DIR}" ] && [ -d "${SYNOLOGY_BACKUP_DIR}/uploads" ]; then
+  echo "Restored uploads mirror from: ${SYNOLOGY_BACKUP_DIR}/uploads"
+elif [ -n "${uploads_file}" ]; then
+  echo "Restored uploads archive: ${uploads_file}"
+fi
+if [ -n "${SYNOLOGY_BACKUP_DIR}" ] && [ -n "${timestamp}" ] && [ -f "${SYNOLOGY_BACKUP_DIR}/config-${timestamp}.tar.gz" ]; then
+  echo "Restored config archive: ${SYNOLOGY_BACKUP_DIR}/config-${timestamp}.tar.gz"
+elif [ -n "${config_file}" ]; then
+  echo "Restored config archive: ${config_file}"
+fi

deploy/cloudflared/config.yml.example

@@ -0,0 +1,14 @@
+ingress:
+  # Primary transcription app endpoint.
+  - hostname: transcription.example.com
+    service: http://app:8000
+
+  # Optional: generic remote access endpoints for other internal services.
+  # Replace hostnames and targets for your LAN.
+  - hostname: homeassistant.example.com
+    service: http://192.168.1.50:8123
+  - hostname: pihole.example.com
+    service: http://192.168.1.60:80
+
+  # Required catch-all.
+  - service: http_status:404

docker-compose.production.yml

@@ -0,0 +1,70 @@
+services:
+  app:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    image: transcription:prod
+    env_file:
+      - .env.production
+    environment:
+      RUN_EMBEDDED_WORKER: "false"
+      RUNTIME_SETTINGS_ENV_FILE: "/app/.env.production"
+    depends_on:
+      postgres:
+        condition: service_healthy
+    ports:
+      - "8000:8000"
+    volumes:
+      - app_uploads:/app/uploads
+      - app_data:/app/data
+      - ./backup:/backup
+      - ./.env.production:/app/.env.production
+      - ./prompts:/app/prompts:ro
+    restart: unless-stopped
+    healthcheck:
+      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/healthz')"]
+      interval: 30s
+      timeout: 5s
+      retries: 5
+      start_period: 20s
+
+  worker:
+    image: transcription:prod
+    env_file:
+      - .env.production
+    command: ["python", "-m", "transcription.worker_service"]
+    depends_on:
+      postgres:
+        condition: service_healthy
+    volumes:
+      - app_uploads:/app/uploads
+      - app_data:/app/data
+      - ./backup:/backup
+      - ./.env.production:/app/.env.production:ro
+      - ./prompts:/app/prompts:ro
+    healthcheck:
+      disable: true
+    restart: unless-stopped
+
+  postgres:
+    image: postgres:16-alpine
+    env_file:
+      - .env.production
+    environment:
+      POSTGRES_DB: ${DATABASE__DATABASE}
+      POSTGRES_USER: ${DATABASE__USER}
+      POSTGRES_PASSWORD: ${DATABASE__PASSWORD}
+    volumes:
+      - postgres_data:/var/lib/postgresql/data
+    restart: unless-stopped
+    healthcheck:
+      test: ["CMD-SHELL", "pg_isready -U $$POSTGRES_USER -d $$POSTGRES_DB"]
+      interval: 10s
+      timeout: 5s
+      retries: 10
+      start_period: 10s
+
+volumes:
+  postgres_data:
+  app_uploads:
+  app_data:

docker-compose.yml

@@ -5,9 +5,12 @@ services:
       dockerfile: Dockerfile
     container_name: transcription-app
     env_file:
-      - .env
+      - .env.production
     environment:
-      DATABASE_URL: sqlite:////app/data/transcription.db
+      # Database configuration uses nested settings names (env_nested_delimiter="__").
+      # DATABASE_URL is NOT read by the application and must not be used here.
+      DATABASE__DRIVER: sqlite
+      DATABASE__PATH: /app/data/transcription.db
       UPLOAD_DIR: /app/uploads
       PROMPT_DIR: /app/prompts
     ports:

docs/Intent.md

@@ -1,24 +0,0 @@
-# Historical Document Transcription Design Intent
-I have several thousand pages of family history told through letters, postcards, books, and other documents that I want to transcribe to text. 
-
----
-
-## Goals
-1. Preserve our family history
-2. Unburden my family (and descendants) from having to store and care for the physical media. Once the documents have been transcribed and organized, they can be donated (or kept by a family member that wants to retain them).
-3. Make the text easily available and searchable by family members, as well as AI (which may have different requirements).
-4. Ability create timelines or assemble the historical record of the family or specific individuals from across the complete document archive. Perhaps use AI to create the timelines in a more narrative form.
-
----
-
-## Source material
-1. **letters, cards, diaries** - handwritten; mostly stored in tubs, with little organization
-2. **books** - typed or typeset; mostly self-published books 50-100 pages in length. This may be expanded to include selected pages from other publications.
-3. **newspaper clippings, event programs, invitations, and other ephemera**
-
----
-
-## Methodology
-
-1. Follow current best practices per "A Guide to Documentary Editing" by Mary-Jo Kline. (See [Transcription Methodology](transcription_methodology.md))
-

docs/architecture.md

@@ -0,0 +1,230 @@
+# System Architecture (Current Baseline: V6.1)
+
+This document defines the current V6.1 architecture baseline.
+
+## Architecture Objectives
+
+- Preserve durable archival records for Documents, Sources, People, and processing runs.
+- Execute page transcription asynchronously with bounded worker behavior.
+- Preserve append-only machine-attempt evidence with request/response provenance.
+- Keep UI, API, service, persistence, and provider boundaries explicit and testable.
+
+## Technical Stack
+
+- **Runtime:** Python 3.12+
+- **Web application:** FastAPI + NiceGUI
+- **Persistence:** SQLModel / SQLAlchemy — PostgreSQL in production, SQLite for local development and tests
+- **Validation and settings:** Pydantic V2 + pydantic-settings
+- **Concurrency:** asyncio worker loop
+- **Provider integration:** OpenRouter adapter behind provider interface
+- **Deployment:** Docker Compose (app, worker, PostgreSQL, Cloudflare Tunnel)
+- **Quality and tests:** Ruff, ty, pytest, pytest-asyncio
+
+## Runtime Topology
+
+```mermaid
+flowchart LR
+U[Browser User] --> A[FastAPI + NiceGUI App]
+A --> W[Asyncio Worker]
+A --> DB[(PostgreSQL / SQLite)]
+W --> P[Provider Adapter]
+W --> DB
+```
+
+The worker loop drains two queues in the same pass: queued transcription Jobs and queued
+`MaintenanceRun` records. When neither has work, it idles.
+
+### Production deployment
+
+Production runs as a Docker Compose stack with the app and worker as separate services, so the app
+process runs with `RUN_EMBEDDED_WORKER=false` and the worker process owns queue draining. Local
+development runs a single process with the worker embedded.
+
+```mermaid
+flowchart LR
+I[Internet] --> CF[cloudflared tunnel + Access]
+CF --> APP[app service]
+APP --> PG[(postgres service)]
+WK[worker service] --> PG
+WK --> PROV[OpenRouter]
+```
+
+Deployment, rollback, and recovery procedures are in [Production Runbook](production-runbook.md);
+backup configuration and restore are in [Backup and Restore](backup_restore.md).
+
+## Layered Boundaries
+
+### Interface Layer
+
+- `src/transcription/ui/**`
+- `src/transcription/api/**`
+
+Responsibilities:
+
+- Route registration, page orchestration, presentation adapters.
+- Structured user messaging through shared error presenter.
+- No direct persistence access from pages/components.
+
+### Service and Orchestration Layer
+
+Aggregate services:
+
+- `src/transcription/services/documents.py`
+- `src/transcription/services/people.py`
+- `src/transcription/services/jobs.py`
+- `src/transcription/services/sources.py`
+- `src/transcription/services/photos.py`
+- `src/transcription/services/maintenance.py`
+- `src/transcription/services/evidence.py` (read/projection only)
+
+Orchestration modules:
+
+- `src/transcription/services/store.py`
+- `src/transcription/services/workflows.py`
+
+Responsibilities:
+
+- Aggregate ownership and invariants.
+- Transaction-aware write helpers.
+- Cross-service workflows in orchestration modules (`store.py`, `workflows.py`).
+- Lookup-table CRUD through the generic `RegistryService` base (`registry.py`), which is not an
+  aggregate owner itself.
+
+Module classification and per-model ownership are defined in
+[services instructions](../.github/instructions/services.instructions.md).
+
+### Persistence Layer
+
+- `src/transcription/db/**`
+
+Responsibilities:
+
+- SQLModel definitions, async session/engine runtime, registry bootstrap.
+- Loader helpers that enforce explicit eager loading with `lazy="raise"` relationships.
+
+### Provider Layer
+
+- `src/transcription/providers/**`
+
+Responsibilities:
+
+- Provider API encapsulation.
+- Request manifest and transport evidence capture.
+- Normalized transcription result contract.
+
+## Core Domain Model
+
+- `Document` owns archival metadata and links to `Source`, `Job`, and `DocumentPerson`.
+- `Source` is a document page/file record with selected machine projection and human revision.
+- `Job` is an aggregate processing run with status and frozen prompt/runtime settings.
+- `JobSource` is queue/membership state for one `(job, source)` pair.
+- `ExecutionAttempt` is append-only evidence for each provider call.
+- `Photo` is person imagery owned by `PhotosService`.
+- `MaintenanceRun` is one queued or executed operational maintenance run.
+- `GenealogyPerson`, `GenealogyFamily`, `GenealogyFamilyChild`, and `GenealogyCitation` store
+  imported GEDCOM genealogy data and citation provenance.
+- `DocumentType` and `PersonRole` are UUID-backed registries with optional protected `semantic_key`.
+- `Tag` is a shared registry reached through both document and person tagging, linked by
+  `DocumentTag` and `PersonTag`.
+
+## Processing and Evidence Workflow
+
+1. User creates/updates Document metadata and linked People atomically through workflow orchestration.
+2. User creates a Job by uploading one or more Source files or by retranscribing an existing Source.
+3. Source files are validated and stored; orientation normalization may be applied at ingest, and stored bytes become the canonical processing bytes.
+4. Worker claims queued Job, transitions to `processing`, and processes pending pages in deterministic order.
+5. Each provider call writes one immutable `ExecutionAttempt` with:
+   - request manifest + hash
+   - transport evidence (when response exists)
+   - SDK snapshot and normalized metadata
+   - outcome, timing, and error details when applicable
+6. `JobSource` status is updated as queue/projection state; `Source.raw_transcription` is set on first successful attempt and can be explicitly re-pointed by candidate promotion.
+7. Job terminal status resolves to `transcribed`, `partial_success`, or `failed`.
+
+## Status Semantics
+
+- **Job statuses:** `queued`, `processing`, `transcribed`, `partial_success`, `failed`
+  - Operational success path resolves to `transcribed`.
+- **JobSource statuses:** `pending`, `transcribed`, `failed`, `cancelled`
+- **MaintenanceRun statuses:** `queued`, `processing`, `succeeded`, `failed`
+  - Maintenance uses `succeeded` rather than `transcribed`; the transcription vocabulary does not
+    apply to operational runs.
+- **Maintenance job types:** `backup`, `storage_reconciliation`, `gedcom_import`
+
+## Maintenance Execution
+
+Operational maintenance is queue-backed rather than run inline from the UI, so it survives request
+lifetime and is recorded:
+
+1. Settings enqueues a `MaintenanceRun` with `status=queued` and a `triggered_by` marker.
+2. The worker claims the oldest queued run with a conditional update, moving it to `processing`.
+3. `backup` runs the deploy backup script; `storage_reconciliation` compares stored media against
+   `Document`/`Source` records; `gedcom_import` parses the latest uploaded `.ged` file and upserts
+   genealogy records.
+4. The run finalizes to `succeeded` or `failed` with summary, timing, log path, and `error_detail`.
+
+`MaintenanceRun` records operational history and is not evidence in the `ExecutionAttempt` sense;
+append-only guarantees apply to transcription attempts.
+
+## Security and Path Handling Boundaries
+
+- Print media delivery uses record-validated API route:
+  - `src/transcription/api/print_api.py`
+- General UI media links resolve through:
+  - `src/transcription/ui/components/media_urls.py`
+- Local filesystem paths must never be accepted from user input as trusted media routes.
+
+## Concurrency and Reliability Principles
+
+- Worker loop reuses service bundle/provider resources for pooled calls.
+- Provider-call timeout is explicit and bounded.
+- Non-retriable worker-loop faults are surfaced and stop loop spin.
+- Per-page outcomes are durably persisted before processing next page.
+
+## Design Decisions and Rationale
+
+### Why `transcribed` is the success terminal state
+
+- The worker and job orchestration resolve successful completion to `JobStatus.TRANSCRIBED`, with mixed and failure outcomes represented by `partial_success` and `failed`.
+- This keeps terminal status vocabulary aligned with what the pipeline actually produces: transcribed page content and evidence, not a generic completion marker.
+
+### Why evidence history is append-only while page text is a projection
+
+- `ExecutionAttempt` stores immutable per-call evidence and preserves full attempt history across retries.
+- `Source.raw_transcription` is intentionally a mutable projection so UI and exports can show a selected current machine text without mutating historical evidence.
+- This split keeps auditability and UX both first-class: history is durable, presentation is editable.
+
+### Why orchestration modules own cross-service workflows
+
+- Service modules do not import each other; aggregate ownership remains local to each service.
+- Multi-aggregate writes are coordinated in orchestration modules (`store.py`, `workflows.py`) so transaction boundaries are explicit and testable.
+- This avoids circular dependencies and keeps cross-cutting workflow logic centralized.
+
+### Why explicit eager loading is required
+
+- ORM relationships are configured with `lazy="raise"` in key paths, so code must request needed relationships up front.
+- This prevents hidden query behavior in UI/service code and makes read shape deterministic and reviewable.
+
+### Why canonical source bytes may be ingest-normalized
+
+- Ingest normalization can correct orientation before persistence so provider calls, evidence hashes, and rendered processing source are consistent.
+- The canonical stored bytes, digest, and size become the durable processing identity for that source.
+
+### Why media access uses controlled routes/helpers
+
+- Print/export media uses record-validated API endpoints to avoid direct filesystem path exposure.
+- General UI media URLs are generated through shared resolver helpers to keep path handling consistent and centralized.
+
+## Scope Boundary
+
+Current architecture rules live in `docs/*`.
+
+## Related References
+
+- [System Requirements](requirements.md)
+- [Data Model](schema.md)
+- [Error Handling Policy](error_handling.md)
+- [Production Runbook](production-runbook.md)
+- [Backup and Restore](backup_restore.md)
+- [Error Handling invariant](./invariant/error_handling.md)
+- [AI evidence invariant](./invariant/ai_evidence_and_provenance.md)

docs/architecture_v2.md

@@ -1,136 +0,0 @@
-# System Architecture (Version 2)
-
-This document describes the V2 production architecture of the personal historical-document transcription system.
-
-## Architecture Objectives
-
-* Preserve source material as immutable transcribed text alongside page-level spatial AI metadata.
-* Support batching multi-image and folder uploads cleanly into sequential pages (`page_number`).
-* Leverage asynchronous worker pools (`asyncio`) for parallel single-image API execution bounded by rate limiters (`asyncio.Semaphore`).
-* Migrate persistence to PostgreSQL using native `UUID`, `TIMESTAMPTZ`, and `JSONB` document storage.
-* Standardize all data validation, API parsing, and database models on **Pydantic V2**.
-* Support rich historical attribution (multi-author and multi-recipient relationships).
-
-## Runtime Topology
-
-The V2 runtime operates as an asynchronous Python application:
-
-* FastAPI + NiceGUI web application process.
-* In-process `asyncio` background task orchestrator for parallel API execution.
-* Relational persistence via PostgreSQL (using `asyncpg` or `psycopg3`).
-* Pydantic V2 validation layer wrapping API payloads and PostgreSQL `JSONB` schemas.
-
-```mermaid
-flowchart LR
-U[Browser User] --> A[FastAPI + NiceGUI App]
-A --> W[Asyncio Worker Engine]
-A --> DB[(PostgreSQL Database)]
-W --> P[Vision Provider APIs\nOpenAI / Claude]
-W --> DB
-```
-
-## Lifecycle Ownership
-
-Application lifespan owns runtime setup/teardown:
-
-* Initialize environment logging and Pydantic configuration.
-* Manage asynchronous PostgreSQL connection pools (`asyncpg` / `psycopg3`).
-* Execute database migrations and index initialization.
-* Recover stale processing jobs on startup.
-* Manage graceful shutdown of active `asyncio` worker pools.
-
-## Layered Module Structure
-
-### Interface Layer
-
-* `src/transcription/ui/**` (NiceGUI pages, multi-page renderers, person cards)
-* `src/transcription/api/**` (FastAPI routes and JSON error handlers)
-
-### Application & Async Worker Layer
-
-* `src/transcription/services/workflows.py`
-* `src/transcription/worker.py`
-
-Responsibilities:
-
-* Batch orchestration and status transitions (`queued` -> `processing` -> `completed` | `partial_success` | `failed`).
-* Parallel single-image API execution using `asyncio.gather` bounded by `asyncio.Semaphore`.
-* Pydantic schema parsing (`PageAIMetadata`) and validation prior to database storage.
-
-### Domain & Service Layer
-
-* `src/transcription/models/*.py` (Pydantic V2 schemas and entity definitions)
-* `src/transcription/services/*.py` (Transactional operations for `Document`, `Person`, `Source`, `Job`, and `JobSource`)
-
-### Infrastructure Layer
-
-* `src/transcription/db/**` (PostgreSQL connection pooling and raw parameterized SQL execution)
-* `src/transcription/providers/**` (OpenAI & Anthropic Vision SDK adapters)
-
-## Processing Workflow
-
-1. User uploads a folder or batch of images for a `Document`.
-2. System creates `Document`, `Job(status='queued')`, and ordered `Source` pages (`page_number = 1..N`).
-3. Worker claims job, sets `Job.status = 'processing'`, and spawns parallel `asyncio` tasks bounded by semaphore.
-4. Each task calls Vision API for a **single** `Source` image.
-5. On task completion:
-* Writes a `JobSource` record containing `status='transcribed'`, `raw_transcription`, `ai_metadata` (bounding boxes/confidence), and `raw_api_response`.
-* Caches active text to `Source.raw_transcription`.
-
-
-6. On page failure:
-* Writes `JobSource` record with `status='failed'` and `error_detail`.
-
-
-7. Once all page tasks resolve:
-* Marks `Job.status` as `completed` (100% success), `partial_success` (at least 1 success, 1 failure), or `failed` (all failed).
-
-
-
-## Domain Ownership & Invariants
-
-* **Immutable AI Outputs:** `source.raw_transcription` and `job_source.raw_transcription` store original, point-in-time machine output and are immutable.
-* **Inlined Revisions:** Human corrections occur on `source.revised_text`. UI renders `COALESCE(revised_text, raw_transcription)`.
-* **Sequential Integrity:** Multi-page documents are strictly ordered by `source.page_number ASC`.
-* **Page Execution Isolation:** A failure on one page image does not invalidate successful transcriptions on sister pages in the same batch job.
-
-## Data Model Summary
-
-* `Document` has many `Source` pages, many `Job` runs, and many `Person` records via `DocumentPerson` junction (`author` or `recipient`).
-* `Source` belongs to one `Document` and can be processed across many `JobSource` executions.
-* `Job` has many `JobSource` execution records.
-
-## Test Strategy
-
-* Unit tests for Pydantic V2 schemas, custom validators, and JSONB serialization.
-* Integration tests for async PostgreSQL connection handling and parameterized queries.
-* Async workflow tests using mock AI providers to verify `partial_success` and retry logic.
-* UI integration tests for multi-page rendering and person management.
-
----
-
-## Technology References
-
-- [FastAPI documentation](https://fastapi.tiangolo.com/)
-- [NiceGUI documentation](https://nicegui.io/documentation)
-- [PostgreSQL documentation](https://www.postgresql.org/docs/)
-- [Python asyncio](https://docs.python.org/3/library/asyncio.html#module-asyncio)
-- [Pydantic Validation](https://pydantic.dev/docs/validation/latest/get-started/)
-- [Pydantic AI](https://pydantic.dev/docs/ai/overview/)
- 
-
-## Related Local References
-
-- [System Overview](index_v1.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- System Architecture (this document)
-- [System Requirements](requirements_v1.md)
-- [Data model](schema_v1.md)
-- [Error Handling Policy](error_handling_v1.md)
-- [Implementation Plan](implementation_plan_v1.md)
-
-
-
-
-

docs/backup_restore.md

@@ -0,0 +1,60 @@
+# Backup and Restore (V6.1)
+
+This guide defines operational backup/restore for clean-slate recovery of the Docker runtime using a host-visible backup folder.
+
+## 1. Backup artifacts
+
+- Backup target root: `BACKUP_DIR` (recommended production value: `/backup`)
+- Database artifact per run:
+  - `postgres-YYYYMMDD-HHMMSS.dump` (PostgreSQL custom dump via `pg_dump -Fc`)
+  - `backup-YYYYMMDD-HHMMSS.manifest` (run manifest)
+- Media/config mirrors under `BACKUP_DIR`:
+  - `uploads/**` (incremental copy: new files only)
+  - `prompts/**` (prompt directory mirror)
+
+Retention:
+
+- `BACKUP_RETENTION_DAYS` applies to `postgres-*.dump` and `backup-*.manifest` files.
+
+## 2. Creating backups
+
+Run from repository root:
+
+```bash
+sh deploy/backup/create_postgres_backup.sh
+```
+
+Environment variables used by the backup script:
+
+- `BACKUP_DIR` (default `./data/backups`)
+- `BACKUP_RETENTION_DAYS` (default `14`)
+- `UPLOAD_DIR` (default `/app/uploads`)
+- `PROMPT_DIR` (default `/app/prompts`)
+- `DATABASE__DRIVER` (must be `postgres`)
+- `DATABASE__HOST` (default `postgres`)
+- `DATABASE__PORT` (default `5432`)
+- `DATABASE__DATABASE` (required)
+- `DATABASE__USER` (required)
+- `DATABASE__PASSWORD` (required)
+
+Recommended production setup:
+
+- Mount a host-visible folder into `/backup` for both `app` and `worker`.
+- Set `BACKUP_DIR=/backup` in `.env.production`.
+- Use host-level tooling (for example Synology Drive Client on the host) to replicate that folder externally.
+
+## 3. Restoring from backup
+
+Restore requires downtime for app + worker writes.
+
+1. Stop app and worker:
+   - `docker compose --env-file .env.production -f docker-compose.production.yml stop app worker`
+2. Restore database:
+   - `sh deploy/backup/restore_postgres_backup.sh /backup/postgres-YYYYMMDD-HHMMSS.dump`
+3. Start app and worker:
+   - `docker compose --env-file .env.production -f docker-compose.production.yml start app worker`
+4. Validate `/healthz` and run one smoke workflow.
+
+Notes:
+
+- `restore_postgres_backup.sh` still supports legacy archive restore paths for older backup sets.

docs/cloudflare_tunnel_access.md

@@ -0,0 +1,66 @@
+# Cloudflare Tunnel and Access Setup
+
+This guide defines the repository-supported setup for exposing app and selected LAN services through Cloudflare Tunnel with Cloudflare Access protection.
+
+## 1. Files used by this deployment
+
+1. `deploy/cloudflared/config.yml` (local copy from `config.yml.example`)
+2. `.env.production` (`CLOUDFLARE_TUNNEL_TOKEN`)
+3. `docker-compose.production.yml` (`cloudflared` service reads token + mounts config)
+
+Do not commit `config.yml` or `.env.production`.
+
+## 2. Configure cloudflared
+
+1. Copy `deploy/cloudflared/config.yml.example` to `deploy/cloudflared/config.yml`.
+2. Update hostname -> service mappings in `ingress`.
+3. Keep the final catch-all ingress `http_status:404`.
+4. Set `CLOUDFLARE_TUNNEL_TOKEN` in `.env.production`.
+
+Example app route:
+
+- `transcription.example.com` -> `http://app:8000`
+
+Optional generic remote-access routes:
+
+- `homeassistant.example.com` -> `http://<home-assistant-lan-ip>:8123`
+- `pihole.example.com` -> `http://<pihole-lan-ip>:80`
+
+## 3. Cloudflare Access policy baseline
+
+Create one Access app policy per exposed hostname:
+
+1. Include: your allowed identities/groups only.
+2. Exclude: none by default.
+3. Require: identity provider login (and MFA if available).
+
+Recommended baseline:
+
+- App endpoint (`transcription.*`): your admin identity set.
+- Other internal endpoints (`homeassistant.*`, `pihole.*`, etc.): explicit least-privilege groups.
+
+## 4. Startup
+
+Start production stack:
+
+```bash
+docker compose --env-file .env.production -f docker-compose.production.yml up -d --build
+```
+
+Validate tunnel container:
+
+```bash
+docker compose --env-file .env.production -f docker-compose.production.yml logs cloudflared
+```
+
+LXC/proxied-network note:
+
+- The `cloudflared` service is pinned to `--protocol http2` with explicit DNS resolvers (`1.1.1.1`, `1.0.0.1`) in `docker-compose.production.yml`.
+- This avoids environments where Docker's embedded resolver (`127.0.0.11`) cannot resolve `region*.v2.argotunnel.com`, which causes connector precheck failure and tunnel shutdown.
+- If tunnel status is still down, verify host/container egress for DNS and TCP 443 to `api.cloudflare.com` and `*.argotunnel.com`.
+
+## 5. Security notes
+
+- Keep `postgres` and other internal-only services off public hostnames unless required.
+- Use distinct hostnames per service; avoid path-based multiplexing for unrelated admin surfaces.
+- Rotate `CLOUDFLARE_TUNNEL_TOKEN` and Access policy memberships on a regular schedule.

docs/data_migration.md

@@ -0,0 +1,94 @@
+# Database Rebuild Migration Workflow
+
+This project uses an explicit **export/import rebuild workflow** for schema migration.
+
+Policy:
+- Do not add runtime legacy-compatibility write paths.
+- Rebuild a fresh target database from current models.
+- Export current data/media, then import into the fresh target.
+
+## Commands
+
+### 1) Export current DB + uploads into a bundle
+
+```bash
+uv run python tools/export_import_migration.py export --bundle-dir .migration-bundle
+```
+
+Optional source overrides:
+- `--source-db <path-or-sqlalchemy-url>`
+- `--source-upload-dir <path>`
+
+### 2) Import bundle into a fresh target (SQLite or PostgreSQL)
+
+```bash
+uv run python tools/export_import_migration.py import --bundle-dir .migration-bundle --target-db .\data\transcription-new.db --target-upload-dir .\data-new
+```
+
+PostgreSQL target example:
+
+```bash
+uv run python tools/export_import_migration.py import --bundle-dir .migration-bundle --target-db postgresql://transcription:change-me@localhost:5432/transcription --target-upload-dir .\data-new
+```
+
+### 3) Verify migration parity and integrity
+
+```bash
+uv run python tools/export_import_migration.py verify --source-db .\data\transcription.db --target-db postgresql://transcription:change-me@localhost:5432/transcription
+```
+
+The verify command checks:
+
+- row-count parity across migration tables
+- orphan-reference checks for `source`, `job`, `job_source`, and `execution_attempt`
+- duplicate `(job_id, source_id, attempt_number)` in `execution_attempt`
+
+Exit code:
+
+- `0` when counts and integrity checks pass
+- `1` when mismatches or integrity violations are detected
+
+### 4) One-shot export+import
+
+```bash
+uv run python tools/export_import_migration.py migrate --bundle-dir .migration-bundle --target-db .\data\transcription-new.db --target-upload-dir .\data-new
+```
+
+## What gets migrated
+
+- Tables (in dependency order): `document_type`, `person_role`, `tag`, `document`, `person`, `photo`, `document_person`, `document_tag`, `person_tag`, `job`, `source`, `job_source`, `execution_attempt`.
+- Media tree under `UPLOAD_DIR`.
+
+The bundle contains:
+- `database.json` (row export)
+- `uploads/` (copied media files)
+
+Path normalization during export/import:
+- `source.file_path` is normalized to `documents/...` (upload-root-relative POSIX).
+- `photo.path` is normalized to `photos/...` (upload-root-relative POSIX).
+
+Legacy V4.x portrait/homepage backfill in the export step:
+- If the source DB has no `photo` table, the exporter synthesizes `photo` rows from legacy `person.portrait_path` values and from legacy homepage image files under `UPLOAD_DIR/homepage`.
+- Legacy portrait and homepage image files are copied into the unified `UPLOAD_DIR/photos/{photo_id}{suffix}` layout in the migration bundle.
+- Legacy homepage markdown is relocated from `UPLOAD_DIR/homepage/homepage.md` to `UPLOAD_DIR/homepage.md`.
+- Legacy `person.full_name` values are split into `given_names` + `last_name` for V5.1 schema compatibility.
+
+## Cutover (SQLite -> PostgreSQL)
+
+After importing to a fresh target:
+1. Stop app and worker services to freeze writes.
+2. Export a migration bundle from the last SQLite state.
+3. Import bundle to PostgreSQL target.
+4. Run `verify` against source and target before switching runtime.
+5. Switch runtime config to PostgreSQL (`DATABASE__DRIVER=postgres` and related `DATABASE__*` values).
+6. Start app and worker services.
+7. Run smoke checks (`/healthz`, create/upload/process one job).
+
+## Rollback
+
+If verify or smoke checks fail:
+
+1. Stop app and worker services.
+2. Revert runtime config to SQLite.
+3. Start app and worker against pre-cutover SQLite database.
+4. Preserve failed migration bundle and logs for analysis.

docs/ddl_v2.sql

@@ -1,102 +0,0 @@
-## PostgreSQL DDL Specification (Version 2)
-
-```sql
--- Enable pgcrypto for UUID generation if on PostgreSQL < 13
-CREATE EXTENSION IF NOT EXISTS "pgcrypto";
-
--- 1. PERSON TABLE
-CREATE TABLE person (
-    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
-    full_name TEXT NOT NULL,
-    display_name TEXT,
-    maiden_name TEXT,
-    birth_date DATE,
-    birth_date_raw TEXT,
-    birth_place TEXT,
-    death_date DATE,
-    death_date_raw TEXT,
-    death_place TEXT,
-    biography TEXT,
-    portrait_path TEXT,
-    metadata JSONB DEFAULT '{}'::jsonb,
-    created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
-    updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
-);
-
--- 2. DOCUMENT TABLE
-CREATE TABLE document (
-    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
-    name TEXT NOT NULL,
-    document_type TEXT,
-    document_date DATE,
-    document_date_raw TEXT,
-    location_created TEXT,
-    notes TEXT,
-    archive_identifier TEXT,
-    created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
-    updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
-);
-
--- 3. DOCUMENT_PERSON (Junction Table for Multi-Author / Multi-Recipient)
-CREATE TABLE document_person (
-    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
-    document_id UUID NOT NULL REFERENCES document(id) ON DELETE CASCADE,
-    person_id UUID NOT NULL REFERENCES person(id) ON DELETE CASCADE,
-    role VARCHAR(20) NOT NULL, -- 'author' or 'recipient'
-    created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
-    CONSTRAINT unique_document_person_role UNIQUE (document_id, person_id, role)
-);
-
--- 4. JOB TABLE (Batch-level orchestrator)
-CREATE TABLE job (
-    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
-    document_id UUID NOT NULL REFERENCES document(id) ON DELETE CASCADE,
-    status VARCHAR(50) NOT NULL DEFAULT 'queued', -- 'queued', 'processing', 'completed', 'partial_success', 'failed'
-    retry_count INTEGER NOT NULL DEFAULT 0,
-    provider TEXT NOT NULL,                        -- e.g., 'openai', 'anthropic'
-    model TEXT NOT NULL,                           -- e.g., 'gpt-4o', 'claude-3-5-sonnet'
-    prompt_name TEXT,
-    date_created TIMESTAMPTZ NOT NULL DEFAULT now(),
-    date_updated TIMESTAMPTZ NOT NULL DEFAULT now()
-);
-
--- 5. SOURCE TABLE (Physical image files & active state)
-CREATE TABLE source (
-    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
-    document_id UUID NOT NULL REFERENCES document(id) ON DELETE CASCADE,
-    page_number INTEGER NOT NULL DEFAULT 1,
-    upload_name TEXT NOT NULL,
-    filename TEXT NOT NULL,
-    file_path TEXT NOT NULL,
-    raw_transcription TEXT,                       -- Cached active AI text output
-    revised_text TEXT,                            -- Active human edited text
-    date_uploaded TIMESTAMPTZ NOT NULL DEFAULT now(),
-    date_revised TIMESTAMPTZ
-);
-
--- 6. JOB_SOURCE (Junction Table: Per-Image Execution Record)
-CREATE TABLE job_source (
-    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
-    job_id UUID NOT NULL REFERENCES job(id) ON DELETE CASCADE,
-    source_id UUID NOT NULL REFERENCES source(id) ON DELETE CASCADE,
-    status VARCHAR(50) NOT NULL DEFAULT 'pending', -- 'pending', 'transcribed', 'failed'
-    raw_transcription TEXT,                       -- Point-in-time raw AI text output
-    ai_metadata JSONB,                            -- Page-level bounding boxes, tokens, confidence
-    raw_api_response JSONB,                       -- Complete REST response envelope
-    error_detail TEXT,
-    executed_at TIMESTAMPTZ NOT NULL DEFAULT now(),
-    CONSTRAINT unique_job_source UNIQUE (job_id, source_id)
-);
-
--- INDEXES FOR FAST LOOKUPS & QUERY PERFORMANCE
-CREATE INDEX idx_person_full_name ON person(full_name);
-CREATE INDEX idx_document_date ON document(document_date);
-CREATE INDEX idx_document_person_doc ON document_person(document_id);
-CREATE INDEX idx_document_person_per ON document_person(person_id);
-CREATE INDEX idx_source_document ON source(document_id);
-CREATE INDEX idx_source_page_order ON source(document_id, page_number);
-CREATE INDEX idx_job_document ON job(document_id);
-CREATE INDEX idx_job_source_job ON job_source(job_id);
-CREATE INDEX idx_job_source_source ON job_source(source_id);
-CREATE INDEX idx_job_source_ai_metadata ON job_source USING GIN (ai_metadata);  
-```

docs/error_handling.md

@@ -0,0 +1,137 @@
+# Error Handling Policy (Current Baseline: V6.1)
+
+This policy defines the active V6.1 error taxonomy, translation boundaries, and retry semantics.
+
+## Error Categories
+
+| Category | Meaning | Typical Origin | User Treatment |
+| :--- | :--- | :--- | :--- |
+| `validation` | Input payload/selection is invalid | UI form parsing, service validators | Inline correction guidance |
+| `not_found` | Target record is missing | ID lookup in service layer | Non-blocking warning or redirect |
+| `conflict` | State prevents requested action | lifecycle transitions, duplicate semantic keys | Explain required precondition |
+| `external` | Provider/network dependency failure | OpenRouter/provider adapter | Retry path and evidence retained |
+| `timeout` | Provider call exceeded configured bound | worker/provider client timeout | Retry path and bounded messaging |
+| `internal` | Unexpected local failure | unhandled service/runtime faults | Safe generic message + diagnostics capture |
+
+## Runtime Taxonomy and Canonical Mapping
+
+Runtime code uses a richer internal taxonomy for diagnostics and persisted evidence, then maps that
+taxonomy to the six canonical categories at the API/UI envelope boundary.
+
+### Internal runtime categories
+
+- `validation_error`
+- `user_input_error`
+- `not_found_error`
+- `conflict_error`
+- `external_provider_error`
+- `external_timeout_error`
+- `processing_error`
+- `infrastructure_transient_error`
+- `infrastructure_persistent_error`
+- `internal_unexpected_error`
+
+### Internal -> Canonical mapping
+
+| Internal category | Canonical envelope category |
+| :--- | :--- |
+| `validation_error` | `validation` |
+| `user_input_error` | `validation` |
+| `not_found_error` | `not_found` |
+| `conflict_error` | `conflict` |
+| `external_provider_error` | `external` |
+| `external_timeout_error` | `timeout` |
+| `infrastructure_transient_error` | `timeout` |
+| `processing_error` | `internal` |
+| `infrastructure_persistent_error` | `internal` |
+| `internal_unexpected_error` | `internal` |
+
+`ExecutionAttempt.error_category` stores the internal category value so diagnostics remain specific.
+
+## Translation Boundaries
+
+- **Provider layer:** raise provider-scoped exceptions with provider context; do not emit UI text.
+- **Service layer:** map raw exceptions into internal categories and preserve causal chain.
+- **UI/API layer:** convert internal categories to canonical categories using the centralized mapping.
+
+## Decision Context
+
+### Why taxonomy is category-based (not exception-class-based)
+
+- Categories encode operator-facing recovery semantics (fix input, retry later, investigate internal failure) independent of low-level exception type.
+- This keeps retry and messaging behavior consistent even when provider/client libraries change.
+
+### Why page-level failure is isolated
+
+- Multi-page archival documents often contain a mix of readable and degraded pages.
+- Isolating failures to page scope preserves successful results and avoids all-or-nothing loss when one page fails.
+- Aggregate job status then communicates overall outcome (`transcribed`, `partial_success`, `failed`) without hiding page detail.
+
+### Why retries append evidence instead of mutating rows
+
+- Retry operations are new observations, not corrections of history.
+- Appending attempts preserves forensic traceability, timing history, and provider variability analysis.
+- Projection updates remain explicit user/workflow decisions, separate from immutable evidence.
+
+## Job and Page Failure Semantics
+
+### Page-Level (`JobSource`)
+
+- `pending` -> `transcribed` when attempt succeeds.
+- `pending` -> `failed` when attempt fails terminally.
+- `pending` -> `cancelled` on job cancellation before processing.
+
+### Job-Level (`Job`)
+
+- `transcribed` when all pages transcribe successfully.
+- `partial_success` when mixed success/failure outcomes exist.
+- `failed` when no page transcribes successfully.
+
+## Retry and Retranscription Rules
+
+1. Failed/cancelled pages may be re-queued through retranscription workflows.
+2. Retry attempts must append new `ExecutionAttempt` rows; prior evidence remains immutable.
+3. Selecting a better candidate must update projection pointers, not mutate historical attempt rows.
+
+## Logging and Diagnostics Rules
+
+1. Persist sufficient attempt error metadata (`error_category`, `error_message`, transport evidence) for post-hoc analysis.
+2. Avoid leaking stack traces or local paths into user-facing message envelopes.
+3. Preserve causal exception chains for internal diagnostics.
+
+### Message vs detail split
+
+Rules 1 and 2 pull in opposite directions: evidence records need the root cause, and
+user-facing envelopes must not carry it. `AppError` therefore separates the two audiences:
+
+| Field | Audience | Carries root cause | Surfaces |
+| --- | --- | --- | --- |
+| `message` | User-facing and API-facing | No | `show_error`, `build_error_envelope` |
+| `detail` | Internal only | Yes | `format_error_detail` (evidence), logs, sanitized UI projection only |
+
+`classify_unexpected_error` builds a generic `message` and puts the exception type and
+text on `detail`. Anything rendered to a user or serialized into an API envelope must
+read `message`; anything persisted as provenance or logged may read `detail`. When a UI
+surface needs to show persisted `error_detail`, it must route through a sanitizing
+projection that preserves the category, suggestion, and error reference while reducing
+machine-local absolute paths to basenames only.
+Enforced by `tests/test_errors.py::test_unexpected_error_does_not_leak_filesystem_paths`
+and `tests/test_error_message_safety.py`.
+
+## Operator Recovery Guidance
+
+- **validation/conflict:** correct input or state and retry manually.
+- **external/timeout:** allow bounded retries and keep prior attempt evidence visible.
+- **internal:** stop automatic retries, surface a safe message, and inspect diagnostics with correlation context.
+
+## UI Messaging Contract
+
+- User-visible errors must be actionable, bounded, and category-consistent.
+- Multi-page jobs must show partial outcomes instead of collapsing into a single opaque failure.
+- Recovery actions (`retry`, `retranscribe`, `edit input`) must be offered where available.
+
+## Cross-Reference
+
+- [Error Handling invariant](./invariant/error_handling.md)
+- [System Requirements](requirements.md)
+- [Data Model](schema.md)

docs/error_handling_v2.md

@@ -1,88 +0,0 @@
-# Error Handling Policy (Version 2)
-
-This document defines the canonical error-handling policy for the V2 document transcription system.
-
-## Error Handling Objectives
-
-* Make failures visible in clear, actionable language at both the document and individual page levels.
-* Support **isolated failure handling** in multi-image batches so single page errors do not crash an entire batch job.
-* Preserve diagnostic detail (Pydantic validation errors, raw provider responses) in PostgreSQL `JSONB` for fast troubleshooting.
-* Ensure consistent error envelope structure across API, UI, and async worker boundaries.
-
-## Scope And Authority
-
-Governs error behavior across NiceGUI pages, FastAPI routes, service orchestration, `asyncio` background tasks, PostgreSQL interactions, and AI provider adapters.
-
-## Error Taxonomy
-
-| Category | Definition | Retriable |
-| --- | --- | --- |
-| `validation_error` | Pydantic payload or parameter schema validation failure | no |
-| `user_input_error` | Unacceptable user file (unsupported image type, corrupt file) | no |
-| `not_found_error` | Requested resource (`Document`, `Source`, `Person`, `Job`) missing | no |
-| `conflict_error` | Operation violates state constraints (e.g., duplicate `document_person` role) | no |
-| `external_provider_error` | AI Provider API failure (rate limit, vision execution error) | yes |
-| `infrastructure_transient_error` | Temporary DB connection reset or HTTP timeout | yes |
-| `infrastructure_persistent_error` | Database down, missing API credentials, misconfiguration | no |
-| `internal_unexpected_error` | Uncaught Python exception or logic defect | no |
-
-## Async Batch & Page-Level Error Behavior
-
-In multi-image `asyncio` batch processing:
-
-1. **Page Isolation:** Exceptions caught during individual page calls are caught within the `asyncio` task wrapper.
-2. **Page Record Logging:** Page failure detail is written directly to `job_source.error_detail` and `job_source.status = 'failed'`.
-3. **Batch Aggregate State:**
-* If **all** page tasks succeed -> `job.status = 'completed'`.
-* If **some** page tasks fail -> `job.status = 'partial_success'`.
-* If **all** page tasks fail -> `job.status = 'failed'`.
-
-
-4. **Retry Strategy:** The UI exposes a "Retry Failed Pages" option for `partial_success` jobs, which spawns a new targeted `Job` containing *only* the `Source` IDs marked as `failed`.
-
-## API Error Response Contract
-
-API error responses return a structured JSON envelope:
-```json
-{
-"error_id": "err_uuid_12345",
-"category": "validation_error",
-"message": "The uploaded payload failed schema validation.",
-"suggestion": "Check file format and metadata fields, then try again.",
-"details": {
-"pydantic_errors": [...]
-},
-"timestamp": "2026-07-31T07:55:00Z"
-}
-```
-
-HTTP Status Mappings:
-
-* `validation_error`, `user_input_error` -> `400`
-* `not_found_error` -> `404`
-* `conflict_error` -> `409`
-* `external_provider_error` -> `502` / `503`
-* `infrastructure_transient_error` -> `503`
-* `infrastructure_persistent_error`, `internal_unexpected_error` -> `500`
-
----
-
-## Technology References
-
-- [FastAPI documentation](https://fastapi.tiangolo.com/)
-- [NiceGUI documentation](https://nicegui.io/documentation)
-- [PostgreSQL documentation](https://www.postgresql.org/docs/)
-- [Python asyncio](https://docs.python.org/3/library/asyncio.html#module-asyncio)
-- [Pydantic Validation](https://pydantic.dev/docs/validation/latest/get-started/)
-- [Pydantic AI](https://pydantic.dev/docs/ai/overview/)
-
-## Related Local References
-
-- [System Overview](index_v2.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v2.md)
-- [System Requirements](requirements_v2.md)
-- [Data model](schema_v2.md)
-- Error Handling Policy (this document)
-- [Implementation Plan](implementation_plan_v2.md)

docs/implementation_plan_v2.md

@@ -1,248 +0,0 @@
-# Implementation Plan (Version 2)
-
-This plan defines the path from the V1 baseline to **Version 2 complete**, aligned to the updated multi-image and multi-person relational domain model:
-
-* `Document` acts as a logical parent container for physical artifacts, supporting multi-author and multi-recipient relationships via `DocumentPerson`.
-* `Source` represents an individual image page within a document, maintaining sequential order (`page_number`), cached active machine output (`raw_transcription`), and inline single user revisions (`revised_text`).
-* `Job` acts as an overarching batch orchestrator for multi-page async processing tasks.
-* `JobSource` records individual point-in-time API executions per image page, storing Pydantic-validated `ai_metadata` and raw REST envelopes (`raw_api_response`).
-* **Pydantic V2** acts as the single source of truth for runtime validation, API payload parsing, and PostgreSQL JSONB serialization.
-
-The objective is to complete the V2 scope with production readiness while keeping non-V2 enhancements out of active delivery.
-
----
-
-## V2 Completion Definition
-
-V2 is complete when all of the following are true:
-
-1. **Functional complete**
-* Multi-image and whole-folder uploads assign sequential page numbers to `Source` records under a single `Document`.
-* Batch jobs process pages concurrently using an `asyncio` worker pool with semaphore rate limiting.
-* Partial job failures resolve cleanly to `partial_success`, allowing single-page retries without re-running successful pages.
-* Multi-author and multi-recipient tagging is supported on `Document`.
-
-
-2. **Data-model complete**
-* SQLite is fully replaced with PostgreSQL (using `asyncpg` or `psycopg3`).
-* Pydantic V2 models validate all API payloads, database row mappings, and `JSONB` structures.
-
-
-3. **Operational complete**
-* Concurrency controls, worker pool metrics, and database connections operate safely under batch load.
-
-
-4. **Documentation complete**
-* `schema_v2.md`, `DDL_v2.sql`, Pydantic model contracts are updated and consistent.
-
-
-
----
-
-## Phase 1 — Data Contract Stabilization & Pydantic Baseline
-
-**Goal:** Lock the PostgreSQL schema, DDL, and Pydantic V2 models before refactoring service logic.
-
-### Tasks
-
-1. Finalize DDL for PostgreSQL native types (`UUID`, `TIMESTAMPTZ`, `JSONB`) and junction tables (`document_person`, `job_source`).
-2. Build core Pydantic V2 schemas (`Person`, `Document`, `Source`, `Job`, `JobSource`, `PageAIMetadata`).
-3. Confirm and document data invariants:
-* `source.raw_transcription` and `job_source.raw_transcription` are immutable machine outputs.
-* `source.revised_text` holds user edits. UI renders `COALESCE(revised_text, raw_transcription)`.
-* Page sequence is strictly ordered by `source.page_number ASC`.
-
-
-4. Freeze V2 job status values (`queued`, `processing`, `completed`, `partial_success`, `failed`) and page execution status values (`pending`, `transcribed`, `failed`).
-
-### Deliverables
-
-* Canonical `docs/schema_v2.md` and `docs/DDL_v2.sql`.
-* Centralized Pydantic validation suite in `models/schemas_v2.py`.
-
-### Exit Criteria
-
-* All database tables, relationships, and JSONB structures have corresponding Pydantic V2 models passing unit validation tests.
-
----
-
-## Phase 2 — Persistence Layer Transition (SQLite to PostgreSQL)
-
-**Goal:** Replace the SQLite storage layer with an asynchronous PostgreSQL driver (`asyncpg` or `psycopg3`).
-
-### Tasks
-
-1. Configure PostgreSQL database connection pooling and environment configuration.
-2. Refactor `services/store.py` / repository layers to execute parameterized async SQL queries (`$1`, `$2`).
-3. Implement JSONB serialization and deserialization helpers using Pydantic's `.model_dump_json()` and `.model_validate()`.
-4. Implement database bootstrap routines for PostgreSQL table creation and index initialization.
-
-### Deliverables
-
-* PostgreSQL-native database connection and query service modules.
-* Integration test suite confirming connection pooling and JSONB CRUD operations.
-
-### Exit Criteria
-
-* All database reads/writes run asynchronously against PostgreSQL with zero remaining SQLite driver dependencies.
-
----
-
-## Phase 3 — Service Layer & `asyncio` Engine Refactor
-
-**Goal:** Implement batch orchestration and parallel single-image API execution.
-
-### Tasks
-
-1. Refactor upload service to process folder/multi-image input:
-* Group files into a single `Document`.
-* Create ordered `Source` rows (`page_number = 1..N`).
-
-
-2. Refactor `services/workflows.py` with `asyncio` worker pools:
-* Use `asyncio.Semaphore` to enforce API provider rate limits.
-* Issue parallel single-image requests to Vision APIs (OpenAI/Claude).
-* Parse API responses directly into Pydantic models (`PageAIMetadata`).
-
-
-3. Update execution tracking:
-* Create a `JobSource` row per page call to record `raw_transcription`, `ai_metadata`, and `raw_api_response`.
-* Update active `source.raw_transcription` upon task completion.
-* Calculate aggregate batch status (`completed`, `partial_success`, `failed`) on the parent `Job`.
-
-
-4. Refactor `services/person.py` and `services/documents.py` to handle multi-person roles via `document_person`.
-
-### Deliverables
-
-* Asynchronous batch execution engine in `services/workflows.py`.
-* Service routines for multi-person tagging and page-level retries.
-
-### Exit Criteria
-
-* Executing a folder upload of 10+ images processes concurrently, populates page-level `JobSource` entries, and handles partial worker errors without crashing the batch.
-
----
-
-## Phase 4 — UI & API Contract Alignment
-
-**Goal:** Update API endpoints and frontend/UI views to render multi-page documents and person roles.
-
-### Tasks
-
-1. Update document and job API endpoints to accept batch file arrays and multi-person ID payloads.
-2. Update UI document views:
-* Render multi-page document transcriptions sequentially by `page_number`.
-* Display author and recipient chips/cards linked from `document_person`.
-
-
-3. Update job detail UI to show page-level execution statuses (`transcribed` vs. `failed`) and provide a "Retry Failed Pages" action for `partial_success` jobs.
-4. Align inline page editing controls to update `source.revised_text` and `source.date_revised`.
-
-### Deliverables
-
-* Refactored API routes and UI components supporting multi-page rendering and person management.
-
-### Exit Criteria
-
-* UI successfully displays multi-page document text, allows per-page human revisions, and shows author/recipient metadata.
-
----
-
-## Phase 5 — Test Suite Realignment & Concurrency Testing
-
-**Goal:** Ensure end-to-end system stability under concurrent async execution and load.
-
-### Tasks
-
-1. Write unit tests for Pydantic models, custom validators, and JSONB conversions.
-2. Write integration tests for async database operations:
-* CRUD for `Document`, `Person`, `DocumentPerson`, `Source`, `Job`, and `JobSource`.
-
-
-3. Write mock-backed async workflow tests:
-* Verify `asyncio.Semaphore` bounds concurrent tasks properly.
-* Validate state transition logic for `completed`, `partial_success`, and `failed` jobs.
-* Confirm retry routines process only targeted `JobSource` records marked as `failed`.
-
-
-4. Re-enable CI quality gates (linting, type checking with Pyright/mypy, pytest).
-
-### Deliverables
-
-* Passing asynchronous test suite covering core workflows, edge cases, and failure recoveries.
-
-### Exit Criteria
-
-* CI pipeline is green with comprehensive coverage across database operations, Pydantic models, and worker queues.
-
----
-
-## Phase 6 — Reliability, Operations, and Release Readiness
-
-**Goal:** Prepare V2 for production deployment and operator management.
-
-### Tasks
-
-1. Verify structured logging includes `job_id`, `document_id`, `source_id`, and `person_id`.
-2. Tune PostgreSQL connection pool limits and `asyncio` concurrency thresholds for production infrastructure.
-3. Update operational documentation:
-* Review and update `docs/schema_v2.md` as needed.
-* Create `docs/runbook_v2.md` detailing PostgreSQL maintenance, JSONB index management, and worker queue monitoring.
-* Create `docs/release_checklist_v2.md` for launch sign-off.
-
-
-
-### Deliverables
-
-* Updated project documentation and operational runbooks.
-* V2 release sign-off checklist.
-
-### Exit Criteria
-
-* All documentation reflects V2 architecture; launch checklist is fully verified.
-
----
-
-## Requirement Traceability Focus
-
-Maintain evidence against these V2 requirement groups:
-
-* **Batch & Multi-Image Pipeline:** Folder ingestion, page ordering, async worker execution.
-* **Database & Persistence:** PostgreSQL, native UUIDs, JSONB execution storage, `asyncpg` pooling.
-* **Validation & Schemas:** Pydantic V2 models for DB rows, API requests, and AI vision responses.
-* **Attribution & Metadata:** Multi-author and multi-recipient tagging, biographical entity management.
-* **Error Recovery:** Partial success states, page-level status flags, isolated retry execution.
-
----
-
-## Scope Discipline Rule (V2 Focus)
-
-* Only tasks required for V2 scope (PostgreSQL, Pydantic V2, folder/async processing, multi-person roles) enter this plan.
-* V3 candidate features (such as side-by-side multi-provider model output comparison) remain strictly in the future backlog.
-* Any schema adjustments during implementation require immediate updates to `DDL_v2.sql`, Pydantic models, and `schema_v2.md`.
-
----
-
-## Technology References
-
-- [FastAPI documentation](https://fastapi.tiangolo.com/)
-- [NiceGUI documentation](https://nicegui.io/documentation)
-- [PostgreSQL documentation](https://www.postgresql.org/docs/)
-- [Python asyncio](https://docs.python.org/3/library/asyncio.html#module-asyncio)
-- [Pydantic Validation](https://pydantic.dev/docs/validation/latest/get-started/)
-- [Pydantic AI](https://pydantic.dev/docs/ai/overview/)
-
-## Related Local References
-
-- [System Overview](index_v2.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v2.md)
-- [System Requirements](requirements_v2.md)
-- [Data model](schema_v2.md)
-- [Error Handling Policy](error_handling_v2.md)
-- Implementation Plan (this document)
- 
-
-

docs/index.md

@@ -0,0 +1,36 @@
+# Document Transcription System Overview (Current Baseline: V6.1)
+
+This directory is the single source of truth for current V6.1 behavior and architecture.
+
+## Canonical Reading Order
+
+1. [System Architecture](architecture.md) for runtime topology, boundaries, and lifecycle ownership.
+2. [System Requirements](requirements.md) for verifiable current-state requirements.
+3. [Data Model](schema.md) for entities, constraints, and evidence persistence rules.
+4. [Error Handling Policy](error_handling.md) for category, translation, and retry behavior.
+
+## Cross-Version Invariants
+
+- [Historical Document Transcription Design Intent](./invariant/intent.md)
+- [Transcription Methodology](./invariant/transcription_methodology.md)
+- [Error Handling](./invariant/error_handling.md)
+- [Digital Evidence and AI Processing Provenance](./invariant/ai_evidence_and_provenance.md)
+- [UI Style Guide](./invariant/ui_style_guide.md)
+
+## Deployment and Operations
+
+- [Production Runbook](production-runbook.md) for deploy, rollback, and recovery.
+- [Backup and Restore](backup_restore.md) for backup configuration and restore procedure.
+- [Data Migration](data_migration.md) for the SQLite to PostgreSQL migration path.
+- [Cloudflare Tunnel and Access](cloudflare_tunnel_access.md) for remote exposure and access control.
+
+## Baseline Statement
+
+The current V6.1 baseline includes the architectural cleanup, person-schema redesign,
+containerized PostgreSQL deployment, and the navigation, Document Detail, and worker-backed
+maintenance refinements reflected across this canonical document set.
+Use this `docs/*` canonical set for active design and implementation decisions.
+
+Every canonical document above states this same baseline; `tests/test_meta_contract_guards.py`
+fails if one of them falls behind. Forward-looking work is tracked in
+[`roadmap_plan.md`](roadmap_plan.md) and is not part of the baseline.

docs/index_v2.md

@@ -1,47 +0,0 @@
-# Document Transcription System Overview (Version 2)
-
-This project is a personal-scale application for transcribing, indexing, and preserving historical family documents, letters, postcards, and journals.
-
-## Start Here
-
-Read [architecture_v2.md](architecture_v2.md) first for technical overview and system design.
-
-## Core V2 Capabilities
-
-* **Folder & Multi-Image Ingestion:** Upload whole folders or image batches that map sequentially (`page_number`) under a single `Document`.
-* **Parallel Async AI Vision Engine:** Concurrently process single-page image transcriptions using Python `asyncio` bounded by rate limiters.
-* **Robust PostgreSQL Storage:** Relational storage for entities with native `UUID`, `TIMESTAMPTZ`, and `JSONB` for deep AI spatial metadata and raw envelopes.
-* **Pydantic V2 Validation:** End-to-end type safety, DB row mapping, and JSONB payload validation.
-* **Historical Person Management:** Track authors and recipients across documents with rich biographical entities (`Person`).
-* **Page-Level Execution Auditing & Revisions:** Store immutable point-in-time machine output per run while enabling inline human corrections (`revised_text`).
-* **Partial Failure Recovery:** Bounded batch execution that isolates single-page API errors (`partial_success`) for simple retries.
-
-## Technical Stack
-
-* **Application Web Framework:** FastAPI + NiceGUI
-* **Persistence Engine:** PostgreSQL 13+
-* **Data Validation & Schemas:** Pydantic V2
-* **Concurrency & Workers:** Python `asyncio` worker pool with `asyncio.Semaphore`
-* **Vision Providers:** OpenAI (GPT-4o) and Anthropic (Claude 3.5 Sonnet) via native SDKs
-
----
-
-## Technology References
-
-- [FastAPI documentation](https://fastapi.tiangolo.com/)
-- [NiceGUI documentation](https://nicegui.io/documentation)
-- [PostgreSQL documentation](https://www.postgresql.org/docs/)
-- [Python asyncio](https://docs.python.org/3/library/asyncio.html#module-asyncio)
-- [Pydantic Validation](https://pydantic.dev/docs/validation/latest/get-started/)
-- [Pydantic AI](https://pydantic.dev/docs/ai/overview/)
- 
-## Documentation Index
-
-- System Overview (this document)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v2.md)
-- [System Requirements](requirements_v2.md)
-- [Data model](schema_v2.md)
-- [Error Handling Policy](error_handling_v2.md)
-- [Implementation Plan](implementation_plan_v2.md)

docs/invariant/ai_evidence_and_provenance.md

@@ -0,0 +1,145 @@
+# Digital Evidence and AI Processing Provenance (Invariant)
+
+## 1. Purpose
+
+This document defines non-negotiable evidence and provenance rules for the transcription application.
+
+The application exists to preserve historical source material and produce useful transcriptions without losing the ability to inspect, reinterpret, or reprocess the evidence later. Provider integrations, model names, schemas, and user interfaces may change; the principles below must remain true.
+
+## 2. Evidence Model
+
+The application distinguishes five kinds of information:
+
+1. **Source evidence**: the canonical stored media used for processing and the facts needed to identify and verify it.
+2. **Execution specification**: the frozen instructions, parameters, source identity, and software context for one processing attempt.
+3. **Transport evidence**: the response received at the application/provider boundary, including safe protocol metadata.
+4. **Normalized data**: selected fields extracted for search, display, accounting, and workflow behavior.
+5. **Derived artifacts**: outputs produced from source evidence, such as transcription text, OCR geometry, confidence data, layout analysis, or entity extraction.
+
+Normalized data and derived artifacts never replace source or transport evidence.
+
+## 3. Core Invariants
+
+### 3.1 Canonical Source Preservation
+
+1. Each source must have one canonical stored byte stream used for processing and provenance.
+2. Canonical storage may apply deterministic ingest normalization before persistence.
+3. Canonical stored bytes must have a cryptographic content digest, byte size, and stable identity.
+4. Post-ingest processing derivatives must not overwrite canonical stored bytes.
+5. Moving or renaming a stored file must not change its evidence identity.
+
+### 3.2 Append-Only Processing History
+
+1. Every processing attempt must have a distinct execution record, whether it succeeds, partially succeeds, times out, or fails.
+2. A later attempt must not overwrite the evidence from an earlier attempt.
+3. A convenient “latest transcription” value may be maintained as a cache or projection, but it is not the authoritative execution history.
+4. Human revisions must remain distinguishable from all machine-generated outputs.
+5. Reprocessing a source must create new evidence rather than rewriting historical evidence.
+
+### 3.3 Frozen Execution Specification
+
+Each execution must preserve enough information to understand what the application asked the processor to do:
+
+1. Requested provider, model, and provider-routing constraints.
+2. Full effective system and user instructions.
+3. Prompt asset name and content digest when a prompt asset is used.
+4. Every explicitly supplied generation or processing parameter.
+5. Whether an optional parameter was explicitly set or omitted.
+6. Canonical source digest (and derivative digests when used), media type, dimensions or page geometry when known, and page identity.
+7. A secret-safe representation of the request structure.
+8. Application, provider-adapter, and client-library versions sufficient to interpret the execution.
+
+The execution specification must not contain credentials, authorization headers, secret query values, or unnecessary duplicate source binaries.
+
+### 3.4 Evidence-Layer Terminology
+
+The following terms are not interchangeable:
+
+- **Transport response**: the status, safe headers, and exact response body received by the application at its HTTP boundary.
+- **Router-normalized response**: a response transformed by an intermediary into its common schema.
+- **SDK-parsed response**: an object created when a client library validates or filters a response.
+- **Normalized metadata**: application-selected fields derived from a response.
+- **Native provider response**: the upstream provider's own response before any intermediary transformation.
+
+The application and its documentation must identify which layer is stored. A response must not be described as “raw,” “complete,” or “native” without naming the boundary at which that claim is true.
+
+### 3.5 Transport Evidence
+
+1. Preserve the exact successful response body received at the application's transport boundary before SDK model parsing can discard unknown fields.
+2. Preserve the response status and an allowlisted set of non-secret headers needed for correlation, content interpretation, rate-limit diagnosis, or audit.
+3. Preserve provider/router request and generation identifiers when available.
+4. Preserve safe response evidence for unsuccessful calls when a response was received.
+5. Record explicitly when no response was received, such as a local timeout or connection failure.
+6. Retain parsed and normalized forms only as additional representations of the preserved response.
+
+Wire-level packet capture, TLS session data, credentials, and unrestricted headers are neither required nor permitted.
+
+These requirements apply to executions performed after transport capture is implemented. For earlier executions, the absence of transport evidence must be represented explicitly. An SDK snapshot or normalized record must never be relabeled or backfilled as transport evidence.
+
+### 3.6 Derived Artifact Provenance
+
+1. Every derived artifact must identify its source evidence and producing execution.
+2. Each artifact must declare its semantic type, media/serialization format, schema name and version, producer, producer version, and creation time.
+3. Artifact content must be stored directly or referenced by a stable path or object identifier and protected by a cryptographic digest.
+4. Coordinates must declare their coordinate system, units, origin, page/image dimensions, and transformation history.
+5. Confidence values must identify the producer and scope to which they apply; values from different producers must not be treated as directly comparable without validation.
+6. Provider-specific payloads may be retained, but durable application behavior must not depend on undocumented provider fields.
+
+This model must accommodate future OCR text, word or line polygons, layout regions, confidence data, alternate transcriptions, and structured extraction without adding a dedicated column for every possible feature.
+
+### 3.7 Integrity and Auditability
+
+1. Stored evidence must be exportable with enough identifiers and metadata to verify relationships and digests outside the application.
+2. Evidence mutation, deletion, and retention behavior must be explicit and testable.
+3. Schema upgrades must preserve existing evidence and its original meaning.
+4. Backfills must be identified as backfills; they must not imply that previously uncaptured evidence existed.
+5. Integrity verification must distinguish a missing file, digest mismatch, unavailable external artifact, and malformed metadata.
+
+### 3.8 Security and Privacy
+
+1. API keys, authorization headers, cookies, and credentials must never be persisted as provenance.
+2. Persist only headers and metadata fields that appear on an explicit allowlist of known-safe fields. Discard all other fields before storage; never persist an unrestricted capture and attempt to redact it afterward.
+3. Request manifests should reference source content by identity instead of duplicating base64 source data.
+4. Diagnostic displays and exports must avoid exposing secrets or machine-local details that are not necessary for evidence interpretation.
+
+## 4. Reproducibility Limits
+
+Provenance supports explanation, comparison, and best-effort reproduction; it does not guarantee identical output.
+
+Identical requests may produce different results because of model updates, provider routing, nondeterministic computation, undocumented defaults, safety systems, or retired endpoints. The application must preserve whether a parameter was omitted rather than pretending to know the provider default used at that time.
+
+Likewise, preserving a general vision-model response does not create OCR coordinates that were never returned. Future coordinate extraction remains possible because canonical source evidence is preserved and can be processed again by a suitable system.
+
+## 5. Model Evaluation Policy
+
+Model selection must be based on a representative sample of the actual archive rather than vendor claims alone.
+
+Evaluation should:
+
+1. Use manually reviewed reference transcriptions following the project's [Transcription Methodology](transcription_methodology.md).
+2. Represent printed, typed, handwritten, degraded, tabular, multilingual, and spatially complex material present in the archive.
+3. Measure character and word error rates where appropriate.
+4. Separately record silent corrections, invented text, omitted text, uncertainty handling, layout fidelity, cost, and latency.
+5. Preserve the exact model, endpoint or route, parameters, prompt, source digest, and scoring method for every comparison.
+6. Treat model rankings as corpus- and version-specific, not permanent declarations of a universal “best” model.
+
+The deterministic scorer for these comparisons lives in `src/transcription/benchmarking.py`; it is
+retained as evaluation-policy infrastructure even though application runtime paths do not call it
+directly.
+
+Benchmark material containing family records remains private application data unless explicitly approved for publication.
+
+## 6. Ownership and Change Policy
+
+1. Canonical V6.1 architecture, schema, requirements, and error-policy documents define how current behavior satisfies this invariant.
+2. Provider adapters own the capture of provider-boundary evidence.
+3. Services own validation, persistence, retention, and export behavior.
+4. UI pages may inspect evidence through service contracts but do not define evidence semantics.
+5. If implementation conflicts with this invariant, either correct the implementation or explicitly revise this document before accepting the behavior.
+6. Revisions to this document require deliberate review because they change the long-term preservation contract.
+
+## 7. Related Invariants
+
+- [Historical Document Transcription Design Intent](intent.md)
+- [Transcription Methodology & Style Guide](transcription_methodology.md)
+- [UI Style Guide](ui_style_guide.md)

docs/invariant/error_handling.md

@@ -0,0 +1,101 @@
+# Error Handling (Invariant)
+
+## 1. Purpose
+
+This document defines the non-negotiable failure-handling principles for the transcription application.
+
+Error categories, API envelopes, status codes, framework integrations, and persistence fields may change between versions. Failures must nevertheless remain visible, safe, diagnosable, and consistent across every application boundary.
+
+## 2. Core Invariants
+
+### 2.1 Failures Are Visible
+
+1. An operation must not report success when all or part of the requested work failed.
+2. Invalid input, unavailable dependencies, persistence failures, provider failures, and unexpected defects must be surfaced through the application's established error path.
+3. Code must not silently discard an exception, provider response, invalid value, or failed state transition.
+4. When work can partially succeed, the successful and failed portions must be identified separately.
+
+### 2.2 Messages Are Actionable
+
+1. Operator-facing errors must explain what failed in concise language.
+2. When a safe corrective action is known, the error must state it.
+3. Expected validation or conflict failures must not be presented as unexplained internal defects.
+4. Internal diagnostics must not replace a usable operator-facing message.
+
+### 2.3 Errors Have Stable Identity and Classification
+
+1. Every surfaced failure must have a stable correlation identifier or equivalent trace identity.
+2. Failures must be classified into a documented, machine-readable category.
+3. Boundary-specific representations must preserve the original category and correlation identity.
+4. Unknown exceptions must be converted at an explicit boundary, retain their causal chain for diagnostics, and be classified as unexpected rather than disguised as an expected failure.
+
+### 2.4 Boundary Translation Is Consistent
+
+1. UI, API, service, worker, persistence, and provider boundaries must use one shared error model or deterministic translations between documented models.
+2. A boundary may simplify presentation, but it must not change the meaning, retryability, or identity of a failure.
+3. Domain and service code must not depend on UI notifications or HTTP response types.
+4. UI and API layers must not infer error categories by parsing message text.
+
+### 2.5 State Changes Are Safe
+
+1. A failed atomic operation must leave persisted state unchanged.
+2. Batch operations may preserve successful independent items only when partial success is an explicit part of the workflow contract.
+3. A failed item must retain enough state to identify what was attempted and whether retry is safe.
+4. Error handling must not overwrite earlier successful results or historical execution evidence.
+
+### 2.6 Retry Is Explicit and Bounded
+
+1. Validation, authorization, policy, conflict, and other deterministic failures must not be retried automatically without a relevant input or state change.
+2. Automatic retry is permitted only for failures classified as transient and only when the operation is idempotent or otherwise protected from duplicate effects.
+3. Retry count, delay, and terminal behavior must be bounded and observable.
+4. Exhausted retries must end in a visible terminal failure rather than an indefinitely pending state.
+
+### 2.7 Diagnostics Are Preserved Safely
+
+1. Logs and persisted diagnostic evidence must retain enough context to correlate the failure with the affected operation and record.
+2. Provider and infrastructure failures must preserve safe diagnostic evidence at the boundary where it is available.
+3. Credentials, authorization headers, cookies, secret values, and unnecessary personal data must not appear in errors, logs, notifications, or exports.
+4. Diagnostic metadata capture must use explicit safe-field allowlists where unrestricted content could contain secrets.
+5. User-facing messages must not expose stack traces, local filesystem details, database credentials, or raw internal exceptions.
+
+AI execution failures also follow the evidence rules in [Digital Evidence and AI Processing Provenance](ai_evidence_and_provenance.md).
+
+### 2.8 Cancellation and Timeout Are Distinct Outcomes
+
+1. User cancellation, application shutdown, local timeout, remote timeout, and provider rejection must remain distinguishable.
+2. Cancellation must not be converted into success or a generic unexpected error.
+3. Timeout handling must identify whether a provider response was received when that fact is known.
+4. Cleanup after cancellation or timeout must preserve consistency and must not conceal a completed side effect.
+
+### 2.9 Logging Must Support Audit Without Becoming the Record
+
+1. Structured logs must include correlation identity, operation, category, and relevant non-secret record identifiers.
+2. Expected operator errors may be logged less severely than unexpected defects, but they must remain observable.
+3. Logs are operational diagnostics and do not replace required database state or archival evidence.
+4. Duplicate logging of the same failure at every layer should be avoided; ownership of the authoritative log event must be clear.
+
+## 3. Verification Policy
+
+Each version must verify:
+
+1. Every documented error category reaches the intended UI and API representation.
+2. Failed atomic writes roll back completely.
+3. Partial-success workflows preserve successful independent results and identify failed items.
+4. Retry behavior is bounded and restricted to eligible failures.
+5. Unexpected exceptions retain correlation and causal information without exposing sensitive details.
+6. Logs, persisted evidence, UI messages, and exports contain no credentials.
+7. Cancellation, timeout, provider response failure, and no-response failure remain distinguishable.
+
+## 4. Versioned Ownership
+
+1. Version-specific error taxonomies, envelopes, HTTP mappings, model fields, and framework behavior belong in the applicable version documentation.
+2. Each versioned error-handling document must state how it satisfies this invariant.
+3. A version may add stricter safeguards but must not weaken these principles without first revising this invariant deliberately.
+4. Implementation and tests must be updated together when a versioned error contract changes.
+
+## 5. Related Invariants
+
+- [Historical Document Transcription Design Intent](intent.md)
+- [Digital Evidence and AI Processing Provenance](ai_evidence_and_provenance.md)
+- [UI Style Guide](ui_style_guide.md)
+

docs/invariant/intent.md

@@ -0,0 +1,25 @@
+# Historical Document Transcription Design Intent
+I have several thousand pages of family history told through letters, postcards, books, and other documents that I want to transcribe to text. 
+
+---
+
+## Goals
+1. Preserve our family history
+2. Unburden my family (and descendants) from having to store and care for the physical media. Once the documents have been transcribed and organized, they can be donated (or kept by a family member that wants to retain and preserve them).
+3. Make the document text easily available and easily searchable.
+4. Ability create timelines for individuals and/or families through document dates or the data contained in them. Perhaps even use AI to generate biographies or family histories.
+
+---
+
+## Source material
+1. **letters, cards, diaries** - handwritten; mostly stored in boxes and tubs with little organization
+2. **books** - typed or typeset; mostly self-published books 50-100 pages in length. This may be expanded to include selected pages from other publications.
+3. **photos** - notes written on the backs of photos and the pages of photo albums
+4. **other ephemera** - newspaper clippings, event programs, invitations, military records, immigration records, etc
+
+---
+
+## Methodology
+
+1. Follow current best practices per **A Guide to Documentary Editing** by Mary-Jo Kline. (See [Transcription Methodology](transcription_methodology.md))
+

docs/invariant/transcription_methodology.md

@@ -45,6 +45,24 @@ The following rules map directly to editorial conventions for handling common ma
 | **Non-Textual Artifacts** | Record non-textual elements (seals, stamps, sketches, physical damage) using brief descriptive text inside square brackets. | [description] | [wax notary seal attached here] or [sketch of a fort layout] |
 | **Marginalia & Addenda** | Explicitly indicate spatial transitions before transcribing content located in margins or non-standard orientations. | [location:] | [written in left margin:] Do not share this with anyone. |
 
+### 3.4 Document-Body Medium
+
+Every transcript must identify the predominant document-body medium exactly once at the beginning:
+
+| Medium | Use | Standard Markup |
+| --- | --- | --- |
+| **Handwritten** | The main body was written by hand. | `[document body handwritten]` |
+| **Typewritten** | The main body was produced with a typewriter. Uneven impressions, monospaced characters, and mechanical defects remain typewritten rather than handwritten. | `[document body typewritten]` |
+| **Typeset** | The main body was composed for printing or produced as printed text rather than with a typewriter. | `[document body typeset]` |
+| **Mixed** | No single medium predominates, or handwritten and printed/typewritten content are structurally interleaved. | `[document body mixed]` |
+
+- Use exactly one document-body marker.
+- Do not wrap each line in `[handwritten: ...]` after declaring the body handwritten.
+- In typewritten or typeset documents, use localized handwriting markers only for genuinely handwritten annotations, insertions, or signatures.
+- In mixed documents, identify handwritten portions locally while preserving their reading context.
+- Preserve tables of contents, tables, forms, columns, captions, marginalia, page numbers, dotted leaders, and associated references in their logical reading order.
+- Produce plain text characters rather than HTML entities for ordinary transcription content.
+
 ## 4. Prompt Asset Integration
 
 When executing programmatic transcriptions via LLM APIs or local models, processing instructions must be packaged into single-purpose system prompts aligned with these rules.

docs/invariant/ui_style_guide.md

@@ -0,0 +1,110 @@
+# UI Style Guide (Invariant)
+
+## 1. Purpose
+This guide defines non-negotiable UI styling rules for the transcription application.
+
+The design system is token-first and class-driven:
+1. Theme tokens are defined in [src/transcription/ui/static/theme.css](../../src/transcription/ui/static/theme.css).
+2. Python UI code composes semantic classes instead of inline color values.
+3. Pages and components should share a single visual language across Documents, Jobs, People, and Sources flows.
+
+## 2. Source of Truth
+Use these files as the style authority:
+1. [src/transcription/ui/static/theme.css](../../src/transcription/ui/static/theme.css) for color tokens, semantic utility classes, table styles, and viewer surfaces.
+2. [src/transcription/ui/theme.py](../../src/transcription/ui/theme.py) for runtime NiceGUI theme bridge and shared UI helpers.
+
+If this document conflicts with implementation, update this document to match the code immediately after intentional style changes.
+
+## 3. Core Design Invariants
+1. Flat, high-density surfaces over decorative depth.
+2. Strong content hierarchy with subdued backgrounds and border-based separation.
+3. Viewer area remains the highest contrast region in image/transcription workflows.
+4. Primary actions are consistent and visually recognizable.
+5. Accessible focus rings are always visible for keyboard users.
+
+## 4. Token System
+
+### 4.1 Palette Tokens
+Base palette variables live under :root in [src/transcription/ui/static/theme.css](../../src/transcription/ui/static/theme.css):
+1. --palette-carbon-black: #1c2321
+2. --palette-cool-steel: #7d98a1
+3. --palette-blue-slate: #5e6572
+4. --palette-powder-blue: #a9b4c2
+5. --palette-platinum: #eef1ef
+
+### 4.2 Semantic Theme Tokens
+Do not style components directly with palette tokens when a semantic token exists.
+
+Semantic tokens currently include:
+1. --theme-text and --theme-text-muted
+2. --theme-page, --theme-surface, --theme-surface-raised, --theme-surface-muted
+3. --theme-border
+4. --theme-primary and --theme-primary-hover
+5. --theme-secondary and --theme-focus
+6. --theme-inverse-text
+7. --theme-viewer, --theme-viewer-border, --theme-viewer-muted
+
+## 5. Approved Semantic Classes
+
+### 5.1 Text and Background
+1. ui-text-primary
+2. ui-text-muted
+3. ui-text-inverse
+4. ui-bg-page
+5. ui-bg-surface
+6. ui-bg-surface-raised
+7. ui-bg-surface-muted
+8. ui-bg-viewer
+9. ui-bg-viewer-overlay
+10. ui-bg-viewer-overlay-soft
+
+### 5.2 Borders and Surfaces
+1. ui-border-subtle
+2. ui-border-viewer
+3. ui-header-divider
+4. ui-card-surface
+5. ui-row-surface
+6. ui-note-box
+7. ui-card-error
+
+### 5.3 Interactive Elements
+1. ui-btn-primary
+2. ui-btn-secondary
+3. ui-link-primary
+4. ui-text-accent
+5. ui-chip-primary
+6. ui-badge-secondary
+7. ui-status and ui-status--<status>
+
+### 5.4 Table Patterns
+1. ui-table
+2. ui-table-header
+3. ui-table-body
+
+Use existing class combinations from [src/transcription/ui/components](../../src/transcription/ui/components) and [src/transcription/ui/pages](../../src/transcription/ui/pages) as reference implementations.
+
+## 6. Legacy Class Policy
+Legacy `vibe-` presentation classes are prohibited. Use `ui-` semantic classes from `theme.css`.
+
+## 7. Prohibited Patterns
+1. Inline hex colors in Python UI class strings or style blocks, except in isolated bridge code explicitly marked for migration.
+2. Ad-hoc one-off class names that duplicate existing semantic class intent.
+3. Page-specific palette forks that bypass theme tokens.
+4. Hidden or low-contrast focus states on interactive controls.
+5. Embedded `<style>` blocks or NiceGUI `.style(...)` calls in Python UI code.
+6. Additional page- or component-specific stylesheets; `theme.css` is the single CSS source.
+
+## 8. Implementation Rules For Contributors
+1. Prefer composing existing semantic classes before creating new ones.
+2. If a new class is required, add it to [src/transcription/ui/static/theme.css](../../src/transcription/ui/static/theme.css) with a semantic name, then reuse it.
+3. Keep behavior ownership in Python and appearance ownership in CSS.
+4. Update UI tests that assert exact text or labels when intentional copy changes are made.
+5. Avoid introducing class churn unrelated to the feature being changed.
+
+## 9. Verification Checklist
+Before merging UI changes, verify:
+1. No new inline hex colors were introduced in UI pages/components.
+2. New styles are token-backed and added to [src/transcription/ui/static/theme.css](../../src/transcription/ui/static/theme.css).
+3. Primary buttons, links, cards, and tables still render with consistent semantics.
+4. Keyboard focus ring visibility is preserved.
+5. Relevant UI and integration tests pass.

docs/production-runbook.md

@@ -0,0 +1,184 @@
+# Production Runbook
+
+This runbook is the operational checklist for releasing and monitoring the transcription system.
+
+## 1. Pre-release gate checklist
+
+1. Run the full suite: `uv run pytest`
+2. Confirm contract guardrails are green:
+   - `uv run pytest tests/test_meta_contract_guards.py`
+3. Confirm health endpoint includes worker liveness payload (`/healthz` returns `worker.state`).
+4. Confirm required runtime settings are present in deployment environment:
+   - `OPENROUTER_API_KEY`
+   - `DATABASE__*`
+   - filesystem paths for data/logs/backups.
+   - `CLOUDFLARE_TUNNEL_TOKEN`
+5. Confirm schema contract alignment is current:
+   - `src/transcription/db/models.py`
+   - `docs/schema.md`
+
+## 2. Release execution steps
+
+1. Deploy artifact/config to target environment.
+   - Production stack: `docker compose -f docker-compose.production.yml up -d --build`
+   - `Settings` loads from explicit `_env_file`, then `ENV_FILE`, then the repository-root `.env.production`; it does not resolve relative to the process working directory.
+   - For Runtime Settings writes in production, mount `.env.production` into the app container and set `RUNTIME_SETTINGS_ENV_FILE=/app/.env.production`.
+   - If deployment uses a non-default env-file location, set both `ENV_FILE` and `RUNTIME_SETTINGS_ENV_FILE` to that absolute path so startup reads and Settings-page writes stay aligned.
+   - For SQLite -> PostgreSQL cutover, run `uv run python tools/export_import_migration.py verify --source-db <sqlite-path-or-url> --target-db <postgres-url>` before switching runtime.
+2. Validate service startup:
+   - `/healthz` responds `200`
+   - if `RUN_EMBEDDED_WORKER=true`, `worker.state` is `running`
+   - if `RUN_EMBEDDED_WORKER=false`, validate `worker` container is running in Compose
+     (worker healthcheck is intentionally disabled because it does not expose `/healthz`)
+   - validate `cloudflared` logs show active tunnel routes and no ingress errors
+3. Execute one smoke workflow:
+   - create a document/job with at least one source
+   - verify terminal job outcome updates
+   - verify execution evidence row appended
+4. Verify log flow:
+   - stdout aggregation receives events
+   - file logs are written under `./data/logs`
+5. Create a fresh PostgreSQL backup after successful deployment:
+   - `sh deploy/backup/create_postgres_backup.sh` (creates DB dump plus uploads/prompts backups under `BACKUP_DIR`)
+
+## 3. Rollback triggers and actions
+
+### Trigger conditions
+
+1. `/healthz` reports `worker.state=failed`
+2. Repeated provider timeout/error spikes beyond normal baseline
+3. Evidence write failures or DB persistence failures
+
+### Actions
+
+1. Roll back app artifact and config to previous release.
+2. Restart service and re-check `/healthz`.
+3. Re-run smoke workflow and confirm worker returns to `running`.
+4. Preserve incident evidence:
+   - `./data/logs`
+   - relevant DB rows (`job`, `job_source`, `execution_attempt`)
+5. If persistence regression is confirmed, restore the latest valid DB dump:
+   - `sh deploy/backup/restore_postgres_backup.sh <dump-file>`
+   - Synology media mirror and paired config snapshot (same timestamp) are restored automatically when present.
+
+## 4. Post-release monitoring checklist
+
+## First 24 hours
+
+1. Monitor `/healthz` periodically for `worker.state`.
+2. Track job terminal distribution (`transcribed`, `partial_success`, `failed`).
+3. Sample timeout/error categories for abnormal increase.
+4. Spot-check new `execution_attempt` records for append-only growth and timing metadata.
+
+## First 72 hours
+
+1. Re-check error/timeout trend versus 24h baseline.
+2. Verify no recurring worker-failed states.
+3. Verify storage growth and rotation behavior under `./data/logs`.
+4. Confirm incident response notes are captured for any production anomalies.
+
+## 5. Operator playbook for common incidents
+
+### Worker failed
+
+1. Check `/healthz` payload (`error_id`, `error_category`).
+2. Locate matching error in logs.
+3. If non-transient defect persists, roll back.
+
+### Provider timeout spike
+
+1. Confirm provider reachability and rate limits.
+2. Review timeout frequency and impacted job volume.
+3. If sustained, execute rollback criteria and notify stakeholders.
+
+### Partial-success increase
+
+1. Inspect affected `job_source` and `execution_attempt` records.
+2. Confirm failures are category-aligned (`external`/`timeout`/`internal`).
+3. Triage whether issue is source quality, provider, or runtime regression.
+
+### Cloudflare ingress/access failure
+
+1. Check `cloudflared` container logs for ingress parse, DNS, or auth failures.
+2. Confirm `deploy/cloudflared/config.yml` hostname mappings are correct.
+3. Confirm `CLOUDFLARE_TUNNEL_TOKEN` in `.env.production` matches the tunnel configured in Cloudflare.
+4. Confirm Cloudflare Access app policy includes the intended identity/group for that hostname.
+5. If logs show SRV/DNS failures via `127.0.0.11`, use the compose-defined resolver override
+   (`dns: 1.1.1.1, 1.0.0.1`) and ensure outbound TCP 443 is allowed.
+
+### Backup or restore failure
+
+1. Verify `postgres` container is healthy and accepting connections.
+2. Confirm dump file exists and is non-zero size.
+3. Re-run backup/restore scripts with explicit `ENV_FILE` and `COMPOSE_FILE` if using non-default paths.
+4. If direct Synology copy fails, keep local backup and resolve mount/network before next backup cycle.
+   - For LXC setups, use `deploy/backup/mount_synology_cifs.example.sh` as the persistent mount template.
+
+## 6. Dependency upgrade policy
+
+Dependencies are declared in `pyproject.toml` and resolved through the committed
+`uv.lock`. The lockfile guarantees reproducible installs; the version specifiers
+control what a deliberate `uv lock --upgrade` is allowed to move.
+
+### NiceGUI is pinned exactly (`nicegui==3.13.0`)
+
+1. **Rationale.** NiceGUI bundles Quasar and Vue. Minor releases change component
+   props, slots, and styling, which surfaces as visual and interaction regressions
+   rather than import or type errors. The UI suite under `tests/ui/` asserts
+   structure and behavior, not rendered appearance, so a NiceGUI bump can pass the
+   full test suite and still degrade the interface.
+2. **Scope of risk.** All NiceGUI usage is confined to `src/transcription/ui/` and
+   uses only the public `nicegui.ui` and `nicegui.events` surfaces. The coupling is
+   shallow, so the pin is about release stability, not about unpicking deep
+   framework entanglement.
+3. **Current stance.** Hold the exact pin through release stabilization. Do not
+   widen it as incidental cleanup, and do not let automated dependency updates move
+   it. This includes forgoing patch releases, which is the accepted cost.
+4. **Revisiting.** Treat a NiceGUI upgrade as scheduled work with its own change
+   window: bump the pin deliberately, run `uv run pytest -m "not external"`, then
+   manually verify each page contract in `docs/ui/pages/` before accepting.
+
+### All other dependencies
+
+Declared with `>=` floors and moved by explicit `uv lock --upgrade`. Verify with
+`uv run ruff check .`, `uv run ty check`, and `uv run pytest -q -m "not external"`
+before committing a changed lockfile.
+
+## 7. Type-check suppression policy
+
+`uv run ty check` is a blocking pre-commit gate. Suppressions are allowed only for
+proven SQLAlchemy descriptor false positives where runtime behavior is correct and
+the checker cannot represent the descriptor protocol at that call site.
+
+Every suppression must be:
+
+1. **Targeted** to a single rule (for example `# ty: ignore[unresolved-attribute]`).
+2. **Inline** on the expression it suppresses (not file-wide).
+3. Followed by a **one-line rationale** stating it is a SQLAlchemy descriptor false positive.
+
+Do not use broad or rationale-free suppressions. If a diagnostic is not a known
+false positive, fix the code instead of suppressing it.
+
+## 8. Worker shutdown budget
+
+Worker shutdown waits for at most:
+
+`WORKER_PROVIDER_TIMEOUT_SECONDS + WORKER_SHUTDOWN_GRACE_SECONDS`
+
+`WORKER_PROVIDER_TIMEOUT_SECONDS` covers an in-flight provider call, and
+`WORKER_SHUTDOWN_GRACE_SECONDS` is extra time for the loop to persist outcomes
+and exit cleanly after the call returns.
+
+Set the container or service termination grace period **above this total**
+budget. If termination grace is shorter, the process may be killed before
+terminal status and evidence writes are finalized.
+
+## 9. Horizontal scaling precondition
+
+Multiple worker replicas can race on execution-attempt numbering for the same
+`(job_id, source_id)` pair. The runtime now retries boundedly on unique-key
+conflicts (`uq_execution_attempt_number`) and surfaces a conflict-domain error
+if retries are exhausted.
+
+Do not deploy additional worker replicas unless this conflict-retry path and its
+tests are present and green in the target build.

docs/requirements.md

@@ -0,0 +1,118 @@
+# System Requirements (Current Baseline: V6.1)
+
+These requirements define the active V6.1 contract and align to current implementation.
+
+Requirement IDs encode the baseline that introduced them (`REQ-4-*` from V4, `REQ-6-*` from V6) and
+are stable. Never renumber an existing ID; retire it explicitly instead.
+
+## Functional Requirements
+
+### Domain and Record Management
+
+- **REQ-4-001 Document Registry:** The system must create and update `Document` records with title, type, date metadata, optional location, optional archive identifier, and optional notes.
+- **REQ-4-002 Source Registry:** The system must create and update `Source` records linked to exactly one `Document`.
+- **REQ-4-003 People Registry:** The system must create and update `Person` records, support many-to-many links to `Document` with role, and support many-to-many Person tagging via the shared Tag registry.
+- **REQ-4-004 Registry Semantics:** Document types and person roles must support optional immutable semantic keys and hard-delete only when unreferenced.
+
+### Job and Workflow Behavior
+
+- **REQ-4-010 Job Creation:** The system must create `Job` records from uploaded sources and from retranscription of existing sources.
+- **REQ-4-011 Prompt Snapshotting:** Job creation must persist effective prompt and runtime settings as immutable per-job snapshots.
+- **REQ-4-012 Queue Membership:** Each `(job, source)` pair must be represented by one `JobSource` row.
+- **REQ-4-013 Job Status Lifecycle:** `Job.status` must use one of `queued`, `processing`, `transcribed`, `partial_success`, `failed`.
+- **REQ-4-014 JobSource Status Lifecycle:** `JobSource.status` must use one of `pending`, `transcribed`, `failed`, `cancelled`.
+- **REQ-4-015 Terminal Job Resolution:** Job terminal status must derive from page outcomes as `transcribed`, `partial_success`, or `failed`.
+- **REQ-4-016 Cancellation Semantics:** Job cancellation must set remaining `pending` page entries to `cancelled`.
+
+### Transcription and Evidence
+
+- **REQ-4-020 Attempt Evidence:** Each provider call must emit one append-only `ExecutionAttempt` record.
+- **REQ-4-021 Attempt Payload:** `ExecutionAttempt` must retain request manifest/hash, outcome, timing, model/provider fields, and error details when present.
+- **REQ-4-022 Transport Evidence:** Provider response evidence must be attached to the attempt when a response is available.
+- **REQ-4-023 Source Projection Rule:** `Source.raw_transcription` is a projection chosen from attempt outcomes and can be repointed by explicit promotion.
+- **REQ-4-024 Candidate Visibility:** UI must expose candidate attempts with metadata needed for comparative review and selection.
+
+### Media and Access
+
+- **REQ-4-030 Ingest Canonicalization:** Stored source bytes may be normalized at ingest (for example orientation correction); stored bytes are the canonical processing source.
+- **REQ-4-031 Path Safety:** Client-facing media URLs must be generated from controlled application paths only.
+- **REQ-4-032 Print Media Validation:** Print/export source media must be served through record-validated API routes.
+
+### Error and UX Contracts
+
+- **REQ-4-040 Error Envelope:** Service/API errors must map to structured, user-safe error categories and messages.
+- **REQ-4-041 Partial Failure Visibility:** Mixed page outcomes must be visible at job and page level.
+- **REQ-4-042 Retry Support:** Failed and cancelled pages must support targeted retranscription without requiring full document recreation.
+
+### Person Imagery
+
+- **REQ-6-001 Photo Records:** The system must store reusable `Photo` records that are either owned by a `Person` or unowned for homepage gallery use.
+- **REQ-6-002 Primary Photo:** At most one photo per owning `Person` may be marked `is_primary`; setting a new primary must clear the previous one.
+- **REQ-6-003 Primary Reassignment:** Deleting an owner's primary photo must promote a remaining photo of that owner rather than leaving the owner without a primary.
+
+### Operational Maintenance
+
+- **REQ-6-010 Queued Maintenance Runs:** Settings-initiated maintenance must persist a `MaintenanceRun` and execute in the worker, not inline in the request that started it.
+- **REQ-6-011 Maintenance Run Types:** `MaintenanceRun.job_type` must use one of `backup`, `storage_reconciliation`, `gedcom_import`.
+- **REQ-6-012 Maintenance Status Lifecycle:** `MaintenanceRun.status` must use one of `queued`, `processing`, `succeeded`, `failed`.
+- **REQ-6-013 Single Claim:** A queued run must be claimed by at most one worker, using a conditional status update rather than read-then-write.
+- **REQ-6-014 Run History:** Completed runs must retain status, timing, summary, log reference, and error detail, and expose the log for viewing and download.
+- **REQ-6-015 GEDCOM Upload Import:** Settings must support manual `.ged` upload and queue-backed import into genealogy tables.
+- **REQ-6-016 GEDCOM Idempotent Upsert:** GEDCOM import must upsert `GenealogyPerson` and `GenealogyFamily` by FamilySearch IDs and avoid duplicate imported citations on re-run.
+
+### Deployment and Runtime Configuration
+
+- **REQ-6-020 Production Persistence:** Production must run against PostgreSQL; SQLite remains supported for local development and tests.
+- **REQ-6-021 Split Worker Deployment:** Production must support running the worker as its own process with the app started at `RUN_EMBEDDED_WORKER=false`.
+- **REQ-6-022 Runtime Settings Persistence:** Runtime settings edits must persist to the mounted production environment file and survive container restart.
+- **REQ-6-023 Configuration Contract Sync:** `.env.production.example` must stay synchronized with `Settings` keys and production-safe defaults.
+- **REQ-6-024 Health Reporting:** The deployed stack must report app and worker health through `/healthz`.
+- **REQ-6-025 Backup and Restore:** Database and media/config backups must be produced on a host-visible path with a tested restore procedure.
+
+## Non-Functional Requirements
+
+- **REQ-4-100 Boundary Integrity:** UI pages/components must not access persistence directly and must call service APIs.
+- **REQ-4-101 Service Ownership:** Aggregate writes must occur in owning service/workflow modules, not in UI handlers.
+- **REQ-4-102 Deterministic Loading:** ORM relationship reads in service/UI code must use explicit eager loading compatible with `lazy="raise"`.
+- **REQ-4-103 Async Safety:** Long-running provider calls must not block UI event handlers directly.
+- **REQ-4-104 Evidence Durability:** Attempt evidence must survive process restart once the transaction commits.
+- **REQ-4-105 Test Guardrails:** Architecture boundary tests must remain in place for services and UI boundaries.
+
+## Requirement Interpretation Notes
+
+### Status and lifecycle semantics
+
+- `REQ-4-013` and `REQ-4-015` intentionally bind success to `transcribed`, not a generic `completed`, so docs, tests, and runtime transitions stay consistent.
+- `REQ-4-016` and `REQ-4-042` distinguish cancellation from failure at page level (`cancelled` vs `failed`) while still allowing targeted retranscription.
+
+### Evidence semantics
+
+- `REQ-4-020` through `REQ-4-024` separate authoritative history (`ExecutionAttempt`) from operational projection (`Source.raw_transcription`).
+- This supports immutable provenance while allowing explicit candidate promotion for operator workflows.
+
+### Boundary and loading semantics
+
+- `REQ-4-100` and `REQ-4-101` codify aggregate/service ownership and keep UI out of persistence concerns.
+- `REQ-4-102` exists to enforce deterministic query shape under `lazy="raise"` and avoid hidden data access in rendering callbacks.
+
+## Verification Anchors
+
+- Service boundary enforcement: `tests/test_service_boundaries.py`
+- UI boundary enforcement: `tests/test_ui_boundaries.py`
+- Job lifecycle reliability and terminal status behavior: `tests/services/test_workflows_reliability.py`
+- Evidence append-only and projection behavior: `tests/services/test_store.py`, `tests/services/test_transcription_service.py`
+- Person photo ownership and primary selection: `tests/services/test_photo_service.py`
+- Maintenance run lifecycle and worker execution: `tests/services/test_maintenance_service.py`
+- Runtime settings persistence: `tests/ui/test_runtime_settings_store.py`, `tests/services/test_settings_services.py`
+- Configuration contract synchronization: `tests/test_meta_contract_guards.py`
+- Deployment health reporting: `tests/api/test_health.py`
+
+## Traceability Notes
+
+- Source of truth for status enums:
+  - `src/transcription/db/models.py`
+- Source of truth for workflow transitions:
+  - `src/transcription/services/workflows.py`
+  - `src/transcription/services/jobs.py`
+- Source of truth for attempt evidence writes:
+  - `src/transcription/services/sources.py`

docs/requirements_v2.md

@@ -1,42 +0,0 @@
-# Document Transcription System Requirements (Version 2)
-
-This document captures the **Version 2 baseline requirements** for the production implementation.
-
-## Requirements Model
-
-| ID | Category | Requirement | Verify Method |
-| --- | --- | --- | --- |
-| REQ-0 | System | Provide end-to-end multi-page document transcription with persistent, inspectable async job states. | demonstration |
-| REQ-1 | Functional | Allow users to upload folders or multi-image batches as sequential `Source` pages under a `Document`. | test |
-| REQ-2 | Functional | Process multi-page jobs asynchronously using an `asyncio` worker pool bounded by rate limits. | test |
-| REQ-3 | Functional | Persist page-level execution outputs (`raw_transcription`, `ai_metadata`, `raw_api_response`) on `JobSource`. | test |
-| REQ-4 | Functional | Support job states (`queued`, `processing`, `completed`, `partial_success`, `failed`) and page states (`pending`, `transcribed`, `failed`). | inspection |
-| REQ-5 | Functional | Allow users to manage historical `Person` records and link multiple authors/recipients to a `Document` via `DocumentPerson`. | test |
-| REQ-6 | Functional | Maintain immutable original machine output on `Source.raw_transcription` while permitting inline human edits on `Source.revised_text`. | test |
-| REQ-7 | Data Constraint | Store all persistent domain data in PostgreSQL using native `UUID`, `TIMESTAMPTZ`, and `JSONB` columns. | inspection |
-| REQ-8 | Data Constraint | Validate all API requests, database rows, and JSONB structures using Pydantic V2 schemas. | test |
-| REQ-9 | Interface | Render multi-page transcriptions sequentially by `page_number` in the web UI with author/recipient metadata. | demonstration |
-| REQ-10 | Operations | Allow operators to retry only failed pages for jobs in a `partial_success` state. | test |
-
-## Element Satisfaction Mapping
-
-* **UI (NiceGUI):** Satisfies REQ-1, REQ-5, REQ-6, REQ-9, REQ-10.
-* **API (FastAPI):** Satisfies REQ-1, REQ-4, REQ-5, REQ-8.
-* **WORKER (asyncio):** Satisfies REQ-2, REQ-3, REQ-4, REQ-10.
-* **PERSISTENCE (PostgreSQL):** Satisfies REQ-3, REQ-6, REQ-7.
-* **MODELS (Pydantic V2):** Satisfies REQ-8.
-
----
-
-## Related Local References
-
-- [System Overview](index_v2.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v2.md)
-- System Requirements (this document)
-- [Data model](schema_v2.md)
-- [Error Handling Policy](error_handling_v2.md)
-- [Implementation Plan](implementation_plan_v2.md)
-
-

docs/reviews/2026-08-23-code-review.md

@@ -0,0 +1,488 @@
+# Architecture & Code Review Report
+
+**Repository Target:** `transcription/`
+**Target Stack:** Python 3.12+ | FastAPI | NiceGUI | SQLModel/SQLAlchemy | Pydantic V2 | asyncio | OpenRouter
+
+**Review date:** 2026-08-23
+**Governing procedure:** `.github/skills/python-code-reviewer/skill.md`
+**Escalations applied:** `.github/skills/evidence-provenance-auditor/skill.md`, `.github/skills/test-effectiveness-auditor/skill.md`
+**Scope:** 77 Python modules / ~13k LOC under `src/transcription`, 57 test files (377 collected non-external tests), 23 documents under `docs/`, 9 active rule files.
+
+> **Status: closed.** Every finding below was remediated in the phases following this
+> review. This document is retained as a record of the reasoning, **not** as a list of
+> open work, and it is not canonical authority.
+>
+> Two recommendations were wrong on contact and were corrected during implementation:
+> the HIGH-03 fix as written would have stripped root-cause data from `ExecutionAttempt`
+> provenance, and the HIGH-01 fix needed to preserve per-page durability that the report
+> did not mention. Where this text and the current code or guard tests disagree, the code
+> and tests are correct.
+
+### Verification commands and outcomes
+
+| Command | Outcome |
+| :--- | :--- |
+| `uv run ruff check .` | **Pass** — `All checks passed!` |
+| `uv run pytest -q -m "not external"` | **Pass** — 377 passed |
+| `uv run ty check` | **10 diagnostics** — all SQLModel/SQLAlchemy column-descriptor false positives (`services/photos.py` ×8, `tests/test_storage_reconciliation.py` ×2). Advisory only; no suppression strategy exists. |
+
+---
+
+## 1. Executive Summary
+
+- **Overall health is good.** The codebase has genuine architectural discipline: layered `ui → services → db`, a single Pydantic-V2 settings source, an atomic compare-and-swap job claim, append-only evidence history, and eleven deterministic guard tests that enforce structural rules rather than describing them.
+- **No Critical findings.** The highest-risk category for this domain — secret leakage into stored provenance — was explicitly audited and **passes**: request headers are never persisted, response headers use an allowlist, and the API key is `SecretStr` end-to-end.
+- **The top risk is a transaction-atomicity violation on the worker hot path.** Page evidence and terminal job status commit in two separate transactions (`workflows.py:549-598`), directly contradicting `services.instructions.md`. A crash between them leaves a transcript persisted against a job stuck in `PROCESSING`.
+- **That violation is invisible to the test suite.** The test-effectiveness audit confirms no test can fail on a split commit — the pipeline tests assert the happy-path end state, which passes either way. The invariant is documented and steered but *not enforced*.
+- **Stale-job recovery is startup-only** (`app.py:79`), with a 30-second staleness threshold. A job orphaned shortly before a fast restart is not recovered and remains `PROCESSING` indefinitely, because the worker only claims `QUEUED` rows.
+- **The mandated error-presentation boundary is bypassed at 8 sites.** `home_page.py` and `people_page.py` hand-roll `ui.notify(str(exc), ...)`, discarding the `error_id`, category, and suggestion that `error_presenter.show_error` provides. `people_page.py` imports the correct helpers and still bypasses them.
+- **User-facing output can leak filesystem paths.** `classify_unexpected_error` (`errors.py:94`) interpolates the raw exception into a message rendered in the UI; a SQLAlchemy `OperationalError` embeds the database file path. This contradicts an explicit rule in `error-handling.instructions.md`.
+- **The retry gate ignores error category** (`workflows.py:185`), so non-retriable faults would be requeued. Currently latent because `worker_max_retries` defaults to `0`.
+- **Highest-leverage work is enforcement, not refactoring.** Two atomicity tests, a `ty` suppression strategy that lets the pre-commit hook become blocking, and `ruff format --check` in the gate would convert three documented-but-unenforced invariants into deterministic ones.
+
+---
+
+## 2. Executive Architecture Assessment
+
+**Verdict: architecturally sound with a concentrated reliability gap in the worker's commit boundary.**
+
+Domain cohesion is strong. The `services/` layer owns transactions and business rules, `ui/` owns presentation, `db/` owns schema, and `providers/` isolates the OpenRouter adapter behind a `TranscriptionProvider` protocol. Dependency direction is correct and — unusually — *mechanically enforced*: `test_service_boundaries.py` AST-scans for service-to-service imports and `test_ui_boundaries.py` scans pages/components for persistence access. Provider details do not leak upward; `workflows.py` imports only the abstract `providers` types, never `openrouter`.
+
+The evidence/provenance model is the strongest part of the system. `ExecutionAttempt` is genuinely append-only, retries append rather than rewrite, projection writes onto `JobSource` are clearly distinguished from history mutation, and all 14 provenance-auditor invariant checks pass.
+
+**Top systemic risks:**
+
+1. **Split commit boundary on the worker path (High).** Evidence durability and job terminal status are two transactions. This is the one place where the architecture's own written contract is contradicted by the implementation, on the hottest path in the system.
+2. **Recovery is a startup-only, time-thresholded sweep (Medium).** There is no runtime reconciliation, so the self-healing property depends on restart cadence rather than on a bounded interval.
+3. **Enforcement coverage has known holes (Medium).** Atomicity, error-presenter usage, and formatting are all documented rules with no deterministic test. The repo's own strength — routing invariants into tests — has not been applied to these three.
+4. **Leaky transaction ownership (Medium).** `workflows.py` reaches into `services.jobs._session_scope()` and `services.sources._session_scope()` — private members of two different services — to open transactions. Session ownership is ambiguous exactly where it most needs to be explicit.
+5. **A 10-diagnostic type-checker baseline with no suppression policy (Low).** The signal is currently ignorable, which means a real regression would blend into the noise.
+
+---
+
+## 3. Findings by Severity
+
+### Critical Severity
+
+**None identified.**
+
+The secret-leakage check — the only plausible Critical for this system — passes explicitly. `OpenRouterProvider` stores an allowlisted subset of *response* headers only (`providers/evidence.py:130-134`, `SAFE_RESPONSE_HEADERS`); request headers containing `Authorization` are never captured into `TransportEvidence`; and the key is held as `SecretStr` from `config.py` through to the client. Append-only evidence history is likewise intact and test-enforced.
+
+---
+
+### High Severity
+
+#### [HIGH-01] Page evidence and terminal job status commit in separate transactions
+
+- **Location:** `src/transcription/services/workflows.py:549-565` (`_finalize_batch_outcome`), `src/transcription/services/workflows.py:584-598` (`_persist_page_outcome`)
+- **Problem & Consequence:** `.github/instructions/services.instructions.md` states: *"Never commit transcript updates separately from the paired terminal/retry job status change."* The implementation does exactly that. `_persist_page_outcome` opens its own scope and commits page evidence (line 592-594); `_finalize_batch_outcome` later opens a *second* scope and commits the terminal `JobStatus` (line 558-560). For a single-page job these are two transactions with a window between them. A process crash, container eviction, or unhandled error in that window persists the transcript while the job remains `PROCESSING`. Because the worker only claims `QUEUED` rows, that job is not reprocessed; it is recoverable only by the startup sweep, and only if it has aged past the staleness threshold (see MED-01). The user sees a job that never completes despite the transcription having succeeded and been billed.
+
+  This is a deliberate design tension, not an oversight: `_persist_page_outcome_durably` (line 568-581) wraps the page write in `asyncio.shield` precisely so per-page evidence survives cancellation mid-batch. That goal is correct for *multi*-page jobs. The defect is that the single-page and final-page cases inherit the split unnecessarily.
+
+- **Recommendation:** Keep per-page durability for intermediate pages, but commit the final page outcome and the terminal status in one transaction.
+
+  ```python
+  # Before — two scopes, two commits
+  await _persist_page_outcome_durably(job=job, services=services, page=page, session=None)
+  ...
+  await _finalize_batch_outcome(job=job, services=services, status=status, session=None)
+
+  # After — final page and terminal status share one transaction
+  async with services.jobs.session_scope() as tx:
+      for page in intermediate_pages:
+          await _persist_page_outcome_durably(job=job, services=services, page=page, session=None)
+      await _write_page_outcome(job=job, services=services, page=final_page, session=tx)
+      await services.jobs.mark_job_status(job.id, status, session=tx)
+      await tx.commit()
+  ```
+
+  Pair this with the atomicity test in HIGH-04 so the boundary cannot silently regress.
+- **Effort:** M
+
+---
+
+#### [HIGH-02] Mandated error-presentation boundary bypassed at 8 sites
+
+- **Location:** `src/transcription/ui/pages/home_page.py:212,220,228,255`; `src/transcription/ui/pages/people_page.py:265,321,330,339`
+- **Problem & Consequence:** `.github/instructions/ui.instructions.md:42` requires all user-facing error display to route through `components/error_presenter.py`. Seven of nine pages comply. These two hand-roll `ui.notify(str(exc), type="negative")`. The consequence is not cosmetic: `show_error` (`error_presenter.py:52-67`) surfaces the correlation `error_id`, the canonical error category, and the actionable `suggestion` field. Bypassing it means a user hitting a failure on the home or people page gets a bare exception string with **no error reference to report**, making these two pages unsupportable in production — precisely the pages most likely to be a user's entry point.
+
+  `people_page.py` already imports `run_ui_action` and `show_error` at lines 28-29 and uses them elsewhere in the same module, so the bypass is inconsistency rather than missing infrastructure.
+- **Recommendation:** Replace each site with the canonical helper. The unused `summarize_error` helper in `error_presenter.py` (currently a retained orphan — see LOW-07) is the natural fit where a compact string is genuinely needed.
+
+  ```python
+  # Before
+  except AppError as exc:
+      ui.notify(str(exc), type="negative")
+
+  # After
+  except AppError as exc:
+      show_error(exc)
+  ```
+
+  Then close the hole permanently by extending `tests/test_ui_boundaries.py` with an AST check that no module under `PAGES_DIR` calls `ui.notify(...)` with `type="negative"`.
+- **Effort:** S
+
+---
+
+#### [HIGH-03] Unexpected-error path leaks filesystem paths into user-facing output
+
+- **Location:** `src/transcription/errors.py:91-98` (line 94), rendered via `src/transcription/ui/components/error_presenter.py:52-67`
+- **Problem & Consequence:** `classify_unexpected_error` builds `f"Unexpected error during {operation}: {exc}"` and stores it as `AppError.message`. `show_error` renders `error.message` directly to the user. Any exception whose `str()` contains infrastructure detail is therefore displayed verbatim — a SQLAlchemy `OperationalError` embeds the absolute SQLite database path, and an `OSError` from the media layer embeds the storage root. `.github/instructions/error-handling.instructions.md:74` states: *"Never leak … local filesystem paths in user-facing output."* This is the generic catch-all path, so it applies to every unanticipated failure across the application.
+- **Recommendation:** Split the diagnostic detail from the user-facing message. Log the full exception with the `error_id` as the correlation key; show the user a stable message plus that id.
+
+  ```python
+  # Before
+  return AppError(
+      f"Unexpected error during {operation}: {exc}",
+      category=ErrorCategory.INTERNAL_UNEXPECTED,
+      ...
+  )
+
+  # After
+  error = AppError(
+      f"Unexpected error during {operation}.",
+      category=ErrorCategory.INTERNAL_UNEXPECTED,
+      suggestion="Retry once. If it persists, report the error reference id.",
+      retriable=False,
+  )
+  logger.exception("error_id=%s operation=%s", error.error_id, operation)
+  return error
+  ```
+
+  Add a case to `tests/ui/test_error_presenter.py` asserting that a raised `OperationalError` carrying a path does not surface that path in the rendered message.
+- **Effort:** S
+
+---
+
+#### [HIGH-04] Transaction-atomicity invariants have no enforcing test
+
+- **Location:** Contract at `.github/instructions/services.instructions.md` §"Workflow Transaction Boundaries"; gap confirmed across `tests/integration/test_pipeline_flow.py:66-160` and `tests/services/test_job_service.py:41-59`
+- **Problem & Consequence:** The test-effectiveness audit establishes that **neither** Transaction B (transcript + `TRANSCRIBED`) nor Transaction C (retry: `error_detail` + `retry_count` + `QUEUED`) is enforced. The existing pipeline test asserts the final state after a successful run — which passes identically whether the writes shared one commit or used two. To fail on a split-commit regression a test must inject a fault *between* the writes; no such test exists.
+
+  The consequence is that HIGH-01 shipped undetected and any future refactor of `advance_job` can reintroduce it just as silently. This is a *governance* failure rather than a code defect: the repo's stated model is that hard rules belong in deterministic tests, and this rule is the most consequential one that never made the transition.
+- **Recommendation:** Add `tests/integration/test_pipeline_atomicity.py` with two tests that patch the session to raise after `flush()` but before `commit()`, then assert that *neither* side of the pair is visible in a fresh session. These tests should **fail against the current implementation** and pass once HIGH-01 is fixed — write them first.
+- **Effort:** M
+
+---
+
+### Medium Severity
+
+#### [MED-01] Stale-job recovery runs only at startup, behind a 30-second threshold
+
+- **Location:** `src/transcription/app.py:71-81` (`_recover_stale_processing_jobs`), sole caller at `app.py:79` inside `_lifespan`
+- **Problem & Consequence:** `requeue_stale_processing_jobs` has exactly one call site, in the lifespan startup handler. There is no runtime re-check. The staleness predicate is `updated_at < now - worker_provider_timeout_seconds` (default **30.0s**, `config.py:116`). A job orphaned less than 30 seconds before a fast container restart therefore fails the predicate at the only moment recovery is attempted, and stays `PROCESSING` forever — the worker claims only `QUEUED` rows. It self-heals only on some *later, unrelated* restart. In a frequently-redeployed environment, restarts are exactly when orphans are created, so the recovery window is systematically misaligned with the failure it exists to handle.
+- **Recommendation:** Move the sweep onto a periodic task in the worker loop (e.g. every `max(30, provider_timeout * 2)` seconds) in addition to the startup call, and derive the threshold from a dedicated `worker_stale_job_seconds` setting rather than reusing the provider timeout, so the two can be tuned independently.
+- **Effort:** M
+
+---
+
+#### [MED-02] Retry gate ignores `error_category`, so non-retriable failures would be requeued
+
+- **Location:** `src/transcription/services/workflows.py:184-194`
+- **Problem & Consequence:** The `JobStatus.FAILED` branch gates solely on `job.retry_count < settings.worker_max_retries`. It does not consult `error_category` or the `AppError.retriable` flag. `.github/instructions/error-handling.instructions.md` classifies `validation`, `not_found`, and `conflict` as non-retriable; under this gate a malformed source or a missing record would be retried to exhaustion, consuming provider quota on calls that cannot succeed and delaying the terminal failure the user needs to see. There is also no backoff — retries requeue immediately.
+
+  Currently **latent**: `worker_max_retries` defaults to `0` (`config.py:113`) and is commented out in `.env`, so the branch always falls through to the max-retries log. It becomes live the moment anyone enables retries.
+- **Recommendation:** Gate on retriability *and* count, and add exponential backoff before requeue.
+
+  ```python
+  case JobStatus.FAILED:
+      if job.error_category in NON_RETRIABLE_CATEGORIES:
+          logger.error("Job %s failed non-retriably (%s).", job.id, job.error_category)
+          return
+      if job.retry_count < settings.worker_max_retries:
+          ...
+  ```
+
+  Cover with a test that a `validation`-category failure is not requeued even when `worker_max_retries > 0`.
+- **Effort:** S
+
+---
+
+#### [MED-03] `IntegrityError` on the attempt-number flush is uncaught, risking evidence loss
+
+- **Location:** `src/transcription/services/sources.py:540-546` (attempt-number computation), `sources.py:587` (unguarded `flush()`)
+- **Problem & Consequence:** `attempt_number` is derived read-then-write as `MAX(attempt_number) + 1`, and `uq_execution_attempt_number` enforces uniqueness (`db/models.py:507`, documented at `docs/schema.md:273`). The sibling `JobSource` insert *does* catch `IntegrityError` (`sources.py:531-534`), but the `ExecutionAttempt` flush at line 587 does not. Two concurrent attempt writes for the same job source would raise an unhandled `IntegrityError` and lose an evidence row — the one class of data this system exists to preserve. Not currently reachable: the worker is single-instance and processes sources sequentially. It becomes reachable the moment a second worker replica is deployed.
+- **Recommendation:** Mirror the `JobSource` handling — catch `IntegrityError`, recompute `MAX(attempt_number) + 1`, and retry the insert a bounded number of times, raising a domain error on exhaustion. Note this constraint as a horizontal-scaling precondition in `docs/production-runbook.md`.
+- **Effort:** M
+
+---
+
+#### [MED-04] Shutdown timeout is shorter than the provider timeout
+
+- **Location:** `src/transcription/worker.py:146` (`asyncio.wait_for(worker_task, timeout=2.0)`); provider timeout at `config.py:116` (default 30.0s)
+- **Problem & Consequence:** Graceful shutdown waits 2 seconds for the worker task, but the stop event is only checked *between* jobs and an in-flight provider call may run for up to 30 seconds. Any shutdown during a provider call therefore cancels mid-flight. Combined with HIGH-01's split commit, a cancellation that lands between the evidence commit and the status commit produces exactly the stuck-`PROCESSING` state described there — so this finding materially raises HIGH-01's probability rather than being independent of it.
+- **Recommendation:** Derive the shutdown budget from the provider timeout (`worker_provider_timeout_seconds + small_grace`) instead of hardcoding `2.0`, and ensure the container's termination grace period exceeds it. Document both in `docs/production-runbook.md`.
+- **Effort:** S
+
+---
+
+#### [MED-05] `workflows.py` reaches into two services' private `_session_scope`
+
+- **Location:** `src/transcription/services/workflows.py:558` (`services.jobs._session_scope()`), `workflows.py:592` (`services.sources._session_scope()`)
+- **Problem & Consequence:** The orchestration module opens transactions by calling a private member on two different service objects. This is the concrete mechanism behind HIGH-01: because transaction ownership is expressed through a private back-door rather than a declared boundary, nothing in the design makes it obvious that two scopes are being opened for one logical unit of work. It also couples `workflows.py` to a service implementation detail that `test_service_boundaries.py` cannot see (it checks imports, not attribute access).
+- **Recommendation:** Promote a single explicit transaction entry point — a `session_scope()` on `ServiceBundle`, or a module-level `unit_of_work(services)` helper — and make `workflows.py` use only that. Extend `test_service_boundaries.py` with an AST check forbidding `_session_scope` attribute access outside the owning service module.
+- **Effort:** M
+
+---
+
+### Low Severity
+
+#### [LOW-01] `hashlib.sha256` over full file bytes runs on the event loop
+- **Location:** `src/transcription/services/store.py:401`
+- **Problem & Consequence:** Digest computation is CPU-bound and synchronous inside an `async def`. For large uploads this blocks the loop, stalling both the NiceGUI UI and the worker. Every sibling I/O path in the codebase correctly uses `asyncio.to_thread` (`media_storage.py:43`, `normalization.py:117`, `photos.py:176`, `sources.py:740,753`), so this is an isolated deviation.
+- **Recommendation:** `digest = await asyncio.to_thread(lambda: hashlib.sha256(file_bytes).hexdigest())`.
+- **Effort:** S
+
+#### [LOW-02] `homepage_store.py` performs synchronous file I/O from async callers
+- **Location:** `src/transcription/ui/homepage_store.py:25,32`; called from `src/transcription/ui/pages/home_page.py:170`
+- **Problem & Consequence:** Same class as LOW-01 — reads/writes the homepage JSON directly rather than via `asyncio.to_thread`. Impact is small (a tiny file), but it is a second deviation from an otherwise universal convention.
+- **Recommendation:** Wrap both calls in `asyncio.to_thread`.
+- **Effort:** S
+
+#### [LOW-03] Worker poll interval is hardcoded outside `Settings`
+- **Location:** `src/transcription/app.py:62` (`poll_interval_seconds=1.0`)
+- **Problem & Consequence:** The single operational knob controlling worker latency-vs-load cannot be tuned without a code change, contradicting the otherwise-clean rule that all configuration lives in `config.py` (zero `os.getenv` calls exist outside it).
+- **Recommendation:** Add `worker_poll_interval_seconds: float = 1.0` to `Settings` and read it at the call site.
+- **Effort:** S
+
+#### [LOW-04] `_build_request_manifest` returns `None` silently, producing incomplete evidence
+- **Location:** `src/transcription/providers/openrouter.py:347`
+- **Problem & Consequence:** When `source_reference is None` the manifest is skipped with no log line. The attempt is still recorded but its provenance is quietly incomplete, and there is no signal that it happened — the failure mode is undetectable after the fact.
+- **Recommendation:** Log at `warning` with the job/source identifiers before returning `None`, so incomplete provenance is at least attributable.
+- **Effort:** S
+
+#### [LOW-05] Ten `ty` diagnostics with no suppression strategy
+- **Location:** `src/transcription/services/photos.py` (8), `tests/test_storage_reconciliation.py` (2)
+- **Problem & Consequence:** All ten are SQLModel/SQLAlchemy false positives — column descriptors are typed as their Python value type (`UUID`, `datetime`, `bool`), so `.is_()`, `.asc()`, `func.count()`, and `group_by()` appear invalid. Because there is no suppression policy, the pre-commit hook must run `ty` in advisory mode, which means a *genuine* new type error would print alongside the known ten and block nothing.
+- **Recommendation:** Add targeted `# ty: ignore[...]` comments with a one-line rationale at each of the ten sites, then flip the pre-commit hook to blocking. This converts a permanently-ignored signal into a real gate.
+- **Effort:** M
+
+#### [LOW-06] `ruff format` is not enforced; 35 files have drifted
+- **Location:** `.pre-commit-config.yaml`, `ruff.toml`
+- **Problem & Consequence:** `ruff check` is blocking but `ruff format --check` is absent from the gate, so formatting drift accumulates silently and inflates unrelated diffs whenever anyone does run the formatter.
+- **Recommendation:** Run `uv run ruff format .` once as a single isolated commit, then add `ruff format --check` to the pre-commit gate.
+- **Effort:** S
+
+#### [LOW-07] Four retained orphans, all recorded as "uncertain — follow-up"
+- **Location:** `tests/test_orphan_sweep.py:33-52` (`KNOWN_ORPHANS`): `BenchmarkManifest`, `dispose_all_engines`, `refresh_engine`, `summarize_error`
+- **Problem & Consequence:** Every entry carries the weakest possible justification. `summarize_error` is the notable one: it is an unused helper in `error_presenter.py` *while two pages hand-roll error display* (HIGH-02) — the orphan and the boundary violation are the same problem viewed from two directions. `dispose_all_engines` / `refresh_engine` are plausibly test-support utilities and should be classified as such rather than left uncertain.
+- **Recommendation:** Resolve each to a definite outcome — `summarize_error` becomes used by the HIGH-02 fix; classify the engine helpers as test-support or delete them; decide on `BenchmarkManifest`.
+- **Effort:** S
+
+#### [LOW-08] Orphan sweep only scans module-level public definitions
+- **Location:** `tests/test_orphan_sweep.py`
+- **Problem & Consequence:** Methods and private functions are out of scope, so dead code inside classes — the most common kind in a service-oriented codebase — is structurally invisible to the sweep.
+- **Recommendation:** Extend the AST walk to public methods on service classes, seeding `KNOWN_ORPHANS` with the current result set to keep the change non-breaking.
+- **Effort:** M
+
+#### [LOW-09] f-string interpolation in logging calls
+- **Location:** `src/transcription/services/workflows.py:193` and similar sites
+- **Problem & Consequence:** `logger.error(f"Job {job.id} has failed...")` formats eagerly regardless of level and prevents structured-logging backends from grouping by template. Ruff's `flake8-logging-format` (`G`) rules are not enabled, so this is unenforced.
+- **Recommendation:** Use `logger.error("Job %s has failed and reached max retries.", job.id)` and enable ruff rule set `G`.
+- **Effort:** S
+
+#### [LOW-10] Low-signal and always-true assertions in the test suite
+- **Location:** `tests/test_traceability.py:54-57`; `tests/integration/test_pipeline_flow.py:135-140,446-452`; `tests/test_orphan_sweep.py:119`; `tests/services/test_workflows_reliability.py:105,178,241,317,375`
+- **Problem & Consequence:** Per the test-effectiveness audit: `test_traceability.py:54-57` asserts properties of dict literals defined in the same file (can only fail if the test itself is edited); `assert processed is True` in the pipeline tests is unfalsifiable because `read_job` raises rather than returning `None`; the `>= 200` orphan threshold is a historical snapshot that tolerates ±40 drift; and the `assert result is not None` guards are shadowed by the attribute assertions that follow. Together these overstate effective coverage.
+- **Recommendation:** Apply the prune/strengthen backlog in §6 (Testing).
+- **Effort:** S
+
+#### [LOW-11] Wall-clock timing dependencies risk CI flakiness
+- **Location:** `tests/services/test_workflows_reliability.py:157-196` (real `time.sleep(0.40)`, upper bound `< 540ms` with only 10% slack); `test_workflows_reliability.py:341` (`asyncio.wait_for(..., timeout=2)`)
+- **Problem & Consequence:** On a loaded CI runner, a 200ms asyncio task plus 400ms blocking setup can exceed the 540ms bound, producing false failures that erode trust in the suite.
+- **Recommendation:** Widen the slack factor to `0.8` or replace the blocking sleep with a controlled clock mock.
+- **Effort:** S
+
+---
+
+## 4. Architectural Drift & Gap Analysis
+
+`Direction` is `doc->code` (implementation must change to match documented intent) or `code->doc` (an undocumented but repeatable convention that should be formalized).
+
+| Area / Component | Direction | Documented / Intended Rule | Actual Implementation State | Severity | Recommended Resolution |
+| :--- | :--- | :--- | :--- | :--- | :--- |
+| Worker commit boundary | `doc->code` | `services.instructions.md`: never commit transcript updates separately from the paired terminal status change | `workflows.py:549-598` commits page evidence and terminal status in two separate sessions | High | Fix per HIGH-01; enforce per HIGH-04 |
+| UI error presentation | `doc->code` | `ui.instructions.md:42`: all user-facing error display routes through `error_presenter.py` | 8 hand-rolled `ui.notify` sites in `home_page.py` and `people_page.py` | High | Fix per HIGH-02; add AST guard to `test_ui_boundaries.py` |
+| Unexpected-error messaging | `doc->code` | `error-handling.instructions.md:74`: never leak local filesystem paths in user-facing output | `errors.py:94` interpolates raw `exc` into the rendered message | High | Fix per HIGH-03 |
+| Retry policy | `doc->code` | `error-handling.instructions.md`: validation / not_found / conflict are non-retriable | `workflows.py:185` gates on retry count only | Medium | Fix per MED-02 |
+| Stale-job recovery | `code->doc` | Not documented as startup-only or time-thresholded | Single startup call site; 30s threshold reuses the provider timeout | Medium | Fix per MED-01, then document the recovery contract in `docs/production-runbook.md` |
+| Transaction ownership | `code->doc` | `services.instructions.md` assigns transaction ownership to services | `workflows.py` opens scopes via two services' private `_session_scope` | Medium | Fix per MED-05; document the single unit-of-work entry point |
+| Blocking-I/O convention | `code->doc` | Not stated as a rule; followed at 5 of 7 sites | `store.py:401` and `homepage_store.py:25,32` deviate | Low | Fix per LOW-01/LOW-02, then state the `asyncio.to_thread` rule in `services.instructions.md` |
+| Configuration centralization | `code->doc` | Zero `os.getenv` outside `config.py` — a real, held convention | Held everywhere except the hardcoded `poll_interval_seconds` at `app.py:62` | Low | Fix per LOW-03, then formalize the rule and add a deterministic guard |
+| Type-check baseline | `code->doc` | No documented policy for `ty` diagnostics | 10 tolerated false positives; hook is advisory-only | Low | Adopt the suppression strategy in LOW-05 and document it |
+| Formatting | `code->doc` | `ruff.toml` configures the formatter | `ruff format --check` absent from the gate; 35 files drifted | Low | Fix per LOW-06 |
+| Dependency pin | — | `docs/production-runbook.md` "Dependency upgrade policy" records the exact `nicegui==3.13.0` pin as a deliberate stability decision | Matches | — | **No action** — correctly documented, not a defect |
+
+---
+
+## 5. Invariant Inventory & Routing Recommendations
+
+| Invariant / Constraint | Current Location | Recommended Target Layer | Rationale |
+| :--- | :--- | :--- | :--- |
+| Transcript + terminal status commit atomically | Instructions only | **Deterministic test** (`tests/integration/test_pipeline_atomicity.py`) | Highest-consequence rule in the system with zero enforcement; steering alone already failed to prevent HIGH-01 |
+| Retry writes commit atomically | Instructions only | **Deterministic test** (same file) | Same class; a partial retry commit corrupts `retry_count` accounting |
+| All UI errors route through `error_presenter` | Instructions (`ui.instructions.md:42`) | **Deterministic test** (extend `test_ui_boundaries.py`) | Mechanically checkable via AST; 8 live violations prove instructions are insufficient here |
+| No filesystem paths in user-facing output | Instructions (`error-handling.instructions.md:74`) | **Deterministic test** (extend `tests/ui/test_error_presenter.py`) | Checkable by asserting a path-bearing exception does not surface its path |
+| Non-retriable categories are never requeued | Instructions | **Deterministic test** (`tests/services/test_workflows_reliability.py`) | Latent today; a test freezes the correct behavior before retries are enabled |
+| Blocking I/O runs via `asyncio.to_thread` | Convention only (5/7 sites) | **Instructions** (`services.instructions.md`) | Judgment-dependent (thresholds vary by payload size); steering fits better than a hard test |
+| Transaction opened through one owned entry point | Convention, violated | **Instructions + test** | Document the entry point; AST-guard against `_session_scope` access outside its owning module |
+| Append-only `ExecutionAttempt` history | Docs + 3 tests | **Keep as-is** | Correctly routed and genuinely mutation-sensitive; the model to imitate |
+| Service/UI boundary rules | Instructions + 2 AST tests | **Keep as-is** | Working exactly as intended |
+| Status vocabulary conformance | `docs/schema.md` + contract guards | **Keep as-is** | Enum drift would fail the suite |
+| No secrets in stored evidence | Docs + provenance skill + allowlist in code | **Keep as-is** | Allowlist is the right mechanism — fails closed by construction |
+| `ty` diagnostic suppression policy | Nonexistent | **Docs + blocking hook** | Needs a written rationale per suppression before the gate can be trusted |
+| NiceGUI exact pin | `docs/production-runbook.md` | **Keep as-is** | Deliberate, documented, correctly excluded from review findings |
+
+---
+
+## 6. Stack-Specific Analysis
+
+### Python 3.12+ Best Practices
+Modern syntax is used consistently: `X | None` unions throughout, builtin generics, no `typing.List`/`Optional` legacy forms, `pathlib` over `os.path`. Type-annotation coverage is high, with no bare `Any` on public service signatures. Broad `except Exception` appears where it belongs — the per-page handler at `workflows.py:352` deliberately isolates one page's failure from the batch, which is correct. `# noqa: PLR0915` / `PLR1702` are used sparingly and consistently. Minor gaps: f-strings in logging (LOW-09), and two blocking-I/O deviations (LOW-01/LOW-02).
+
+### FastAPI
+Lifespan is handled correctly via an `asynccontextmanager` `_lifespan` (`app.py:36-68`) rather than deprecated `@app.on_event`. Routers are domain-organized with typed path/query parameters and `response_model` declarations. Error handling is centralized through `register_error_handlers`, and the full internal→canonical category mapping is round-trip tested at the HTTP layer (`tests/api/test_error_responses.py:59-95`). `print_api.py:42-49` performs correct `relative_to`-based path containment for media serving. No blocking calls found in `async def` route handlers.
+
+### NiceGUI
+Separation of concerns is good — pages delegate to services and `test_ui_boundaries.py` mechanically prevents persistence access from pages and components. Client state is client-scoped; no cross-session global-state leaks found. API usage is correct for the pinned 3.13.0 release. The two defects are the error-presenter bypass (HIGH-02) and synchronous file I/O in `homepage_store.py` (LOW-02).
+
+### SQLModel & SQLAlchemy
+The strongest layer. `lazy="raise"` is declared on relationships and correctly paired with `expire_on_commit=False`, which together make N+1 access a loud failure rather than a silent performance cost — no N+1 patterns found. The job claim is a genuine atomic compare-and-swap (`jobs.py:212-222`: conditional `UPDATE ... WHERE status = QUEUED ... RETURNING`), which is the correct primitive and correctly implemented. Hot-path indexes are declared and test-verified (`test_db.py:131`). Cross-dialect portability is handled for SQLite and PostgreSQL. Weaknesses are transaction *ownership* (MED-05, HIGH-01) rather than query construction, plus the uncaught `IntegrityError` at MED-03.
+
+### Pydantic V2 & Settings
+Fully migrated — no `@validator`, no `Config` class, no `.dict()` or `parse_obj` anywhere. `model_config = ConfigDict(...)` and `@field_validator` are used correctly. `config.py` is a clean single source of truth: **zero** `os.getenv` calls exist outside it, `.env` is untracked and gitignored, and the API key is `SecretStr` end-to-end. The only deviation is the hardcoded poll interval (LOW-03).
+
+### Asyncio Workers
+Task lifecycle is handled properly: task references are retained (no GC risk), `CancelledError` is re-raised rather than swallowed, the provider call happens outside any DB transaction, timeouts resolve to terminal states, and there is no tight polling spin. `_persist_page_outcome_durably`'s use of `asyncio.shield` (`workflows.py:568-581`) is a thoughtful durability mechanism. The defects are the split commit boundary (HIGH-01), the shutdown-vs-provider timeout mismatch (MED-04), and startup-only recovery (MED-01).
+
+### OpenRouter / Adapter Boundary
+Encapsulation is clean — `workflows.py` imports only abstract types from `providers`, never `openrouter` directly, so provider specifics do not leak into business logic. The `AsyncClient` is shared with configured timeouts and is properly closed: `worker.py:248,271` → `services.aclose()` → `sources.aclose()` (`sources.py:129-133`) → provider `aclose()` (`openrouter.py:86-87,233-235`). Responses are Pydantic-validated. **All 14 evidence-provenance-auditor invariant checks pass**, including the critical one: the API key is never persisted, request headers are never stored, and `TransportEvidence` captures response headers through an explicit allowlist (`evidence.py:130-134`). Only LOW-04 applies here.
+
+### Testing & Quality Tooling
+377 tests pass with `-m "not external"`. The project test contract is honored: `--strict-markers` with all three markers (`unit`, `integration`, `external`) declared, `asyncio_mode = "strict"` with **every** `async def test_` correctly decorated across all 17 async test files, `external` properly excluded from default runs, and **no unawaited-coroutine warnings** — the `filterwarnings` error promotion is clean.
+
+Contract coverage is genuinely strong for structural rules. Confirmed *mutation-sensitive* enforcement exists for: append-only evidence history (3 independent tests, including full before/after field-tuple snapshots), stuck-in-`PROCESSING` prevention, the complete 10-category error mapping, and both boundary rules.
+
+The critical gap is transaction atomicity (HIGH-04) — the audit verdict is **"Effective with Conditions / Go with Conditions"**, blocking on the two missing atomicity tests. Secondary items are the low-signal assertions (LOW-10) and wall-clock flakiness (LOW-11).
+
+**Prune/strengthen backlog:**
+
+| Priority | Task | Location |
+| :--- | :--- | :--- |
+| High | Add Transaction B atomicity test (fault injected between transcript and status writes) | new `tests/integration/test_pipeline_atomicity.py` |
+| High | Add Transaction C atomicity test (retry: `error_detail` + `retry_count` + `QUEUED`) | same file |
+| Medium | Delete tautological assertions on same-file dict literals | `tests/test_traceability.py:54-57` |
+| Medium | Remove unfalsifiable `assert processed is True` | `tests/integration/test_pipeline_flow.py:135-140,446-452` |
+| Medium | Replace `>= 200` snapshot threshold with set-membership assertion | `tests/test_orphan_sweep.py:119` |
+| Medium | Assert mapped test files contain ≥1 test, not merely that they exist | `tests/test_traceability.py:59-60` |
+| Low | Widen timing slack or mock the clock | `tests/services/test_workflows_reliability.py:157-196` |
+| Low | Drop `assert result is not None` guards shadowed by following assertions | `tests/services/test_workflows_reliability.py:105,178,241,317,375` |
+
+---
+
+## 7. Duplication & Consolidation Report
+
+| Pattern / Duplication | Locations | Proposed Canonical Home | Estimated Lines Removed |
+| :--- | :--- | :--- | :--- |
+| Hand-rolled `ui.notify(str(exc), type="negative")` | `home_page.py:212,220,228,255`; `people_page.py:265,321,330,339` | `ui/components/error_presenter.py::show_error` (already exists) | ~16 |
+| Optional-session `if session is None: async with _session_scope()` preamble | `workflows.py:557-561`, `workflows.py:591-595`, and sibling service write paths | `services/base.py::unit_of_work(services, session)` context manager | ~30 |
+| Synchronous I/O not wrapped in `asyncio.to_thread` | `store.py:401`, `homepage_store.py:25,32` | `services/base.py::run_blocking` helper | ~6 |
+| Read-then-increment `MAX(n) + 1` with uniqueness retry | `sources.py:540-546` (uncaught) vs `sources.py:531-534` (caught) | `services/base.py::insert_with_sequence_retry` | ~20 |
+
+### Proposed Canonical Abstractions
+
+```python
+# src/transcription/services/base.py
+
+
+@asynccontextmanager
+async def unit_of_work(
+    services: ServiceBundle,
+    session: AsyncSession | None = None,
+) -> AsyncIterator[AsyncSession]:
+    """Single transaction entry point. Yields a session and commits once on clean exit.
+
+    Replaces the `if session is None: async with X._session_scope()` preamble and the
+    private-member access at workflows.py:558,592. Makes the two-commit split of
+    HIGH-01 structurally hard to reintroduce.
+    """
+
+
+async def run_blocking[T](fn: Callable[[], T]) -> T:
+    """Run a CPU- or disk-bound callable off the event loop."""
+    return await asyncio.to_thread(fn)
+
+
+async def insert_with_sequence_retry(
+    session: AsyncSession,
+    *,
+    build: Callable[[int], SQLModel],
+    next_value: Callable[[], Awaitable[int]],
+    attempts: int = 3,
+) -> SQLModel:
+    """Insert a row carrying a derived sequence number, retrying on IntegrityError."""
+```
+
+---
+
+## 8. Meta-Tooling & Instruction Update Recommendations
+
+1. **Add `tests/integration/test_pipeline_atomicity.py`** (HIGH-04). The single highest-value enforcement change. Write it before fixing HIGH-01 so it demonstrably fails first.
+2. **Extend `tests/test_ui_boundaries.py`** with an AST check forbidding `ui.notify(..., type="negative")` in `PAGES_DIR`, routing all error display through `error_presenter`. Converts `ui.instructions.md:42` from steering into enforcement.
+3. **Extend `tests/ui/test_error_presenter.py`** with a case asserting that a path-bearing exception does not surface its path, enforcing `error-handling.instructions.md:74`.
+4. **Adopt a `ty` suppression policy** — targeted `# ty: ignore[...]` with rationale at the 10 known sites, documented in `docs/` — then **flip the pre-commit `ty` hook from advisory to blocking**. Until this happens the type checker provides no gate.
+5. **Add `ruff format --check` to the pre-commit gate**, preceded by one isolated formatting commit across the 35 drifted files.
+6. **Enable ruff rule set `G`** (`flake8-logging-format`) to catch f-string logging (LOW-09).
+7. **Extend `tests/test_orphan_sweep.py`** to public methods on service classes, seeding `KNOWN_ORPHANS` with current results (LOW-08). Then resolve all four existing "uncertain" entries to definite outcomes.
+8. **Extend `tests/test_service_boundaries.py`** with an AST check forbidding `_session_scope` attribute access outside its owning service module (MED-05). Also address the noted classification gap: the test excludes orchestration modules by hardcoded stem name (`store`, `workflows`, `__init__`), so a new orchestration module under a different name would be misclassified as a service.
+9. **Update `.github/instructions/services.instructions.md`** to state the `asyncio.to_thread` rule for blocking I/O and to name the single `unit_of_work` transaction entry point.
+10. **Update `docs/production-runbook.md`** with the stale-job recovery contract (interval, threshold, and its relationship to the container termination grace period), and note single-worker as a current precondition until MED-03 is fixed.
+11. **Note for `test_ui_boundaries.py`:** the forbidden-import lists are fixed string sets, so a future persistence helper under a new name would escape the check. Consider inverting to an allowlist of permitted imports for pages.
+
+---
+
+## 9. Prioritized Dependency-Ordered Action Plan
+
+**Phase 1: Blocking fixes**
+1. Write the two atomicity tests (HIGH-04) and confirm they **fail** against current `main`.
+2. Fix the split commit boundary (HIGH-01) and confirm the tests now pass.
+3. Fix the filesystem-path leak in `classify_unexpected_error` (HIGH-03).
+4. Replace the 8 hand-rolled error notifications with `show_error` (HIGH-02).
+
+**Phase 2: Enforcement hardening**
+5. Add the `ui.notify` AST guard and the path-leak presenter test, locking in items 3-4.
+6. Adopt the `ty` suppression policy and make the pre-commit hook blocking (LOW-05).
+7. Run `ruff format .` as an isolated commit, then add `ruff format --check` to the gate (LOW-06).
+8. Enable ruff rule set `G` and fix the resulting logging call sites (LOW-09).
+
+**Phase 3: Reliability & concurrency**
+9. Move stale-job recovery to a periodic worker task with a dedicated setting (MED-01).
+10. Gate retries on `error_category` and add backoff (MED-02) — do this before ever raising `worker_max_retries` above 0.
+11. Derive the shutdown budget from the provider timeout (MED-04).
+12. Handle `IntegrityError` on the attempt-number flush (MED-03) — a hard precondition for running more than one worker replica.
+13. Move `sha256` and homepage-store I/O off the event loop (LOW-01, LOW-02); move the poll interval into `Settings` (LOW-03).
+
+**Phase 4: Consolidation & refactoring**
+14. Introduce `unit_of_work` and migrate `workflows.py` off private `_session_scope` access (MED-05); add the corresponding boundary guard.
+15. Extract `run_blocking` and `insert_with_sequence_retry` (§7).
+16. Prune the low-signal assertions and reduce timing flakiness (LOW-10, LOW-11).
+
+**Phase 5: Non-blocking governance/documentation depth**
+17. Extend the orphan sweep to methods and resolve the four uncertain orphans (LOW-07, LOW-08).
+18. Update `services.instructions.md` and `docs/production-runbook.md` per §8 items 9-10.
+19. Log incomplete request manifests (LOW-04).
+20. Consider inverting the UI boundary check to an allowlist.
+
+---
+
+## 10. Preserved Strengths
+
+- **Evidence and provenance integrity is exemplary.** All 14 provenance-auditor invariants pass. `ExecutionAttempt` history is genuinely append-only, retries append rather than rewrite, and projection writes are cleanly distinguished from history mutation. Three independent tests — including full before/after field-tuple snapshots — make any mutation regression fail loudly.
+- **Secret hygiene is correct by construction.** The response-header **allowlist** (`evidence.py:130-134`) fails closed: a newly-introduced sensitive header is excluded by default rather than requiring someone to remember to block it. Request headers are never captured, and `SecretStr` is used end-to-end.
+- **Atomic job claiming.** `jobs.py:212-222` uses a conditional `UPDATE ... WHERE status = QUEUED ... RETURNING` — a true compare-and-swap that makes double-claiming impossible under concurrency, rather than the common read-then-write race.
+- **`lazy="raise"` paired with `expire_on_commit=False`.** This combination turns accidental lazy loads into immediate errors instead of silent N+1 queries, and it is the reason no N+1 patterns exist in the codebase. Keep it.
+- **Architectural rules are mechanically enforced, not merely documented.** AST-based boundary tests for service-to-service imports and UI persistence access are the right pattern; this review's main recommendation is simply to apply that same pattern to three more rules.
+- **Configuration discipline.** Zero `os.getenv` calls outside `config.py`, `.env` untracked and gitignored, clean Pydantic V2 throughout with no V1 residue.
+- **Path containment on media serving.** `print_api.py:42-49` uses proper `relative_to` validation rather than string prefix matching.
+- **Async worker fundamentals.** Task references retained, `CancelledError` re-raised, provider calls outside DB transactions, timeouts resolving to terminal states, no tight polling loop. `asyncio.shield` in `_persist_page_outcome_durably` is a genuinely thoughtful durability mechanism — the fix in HIGH-01 should preserve it for intermediate pages.
+- **Test contract rigor.** `--strict-markers`, `asyncio_mode = "strict"` honored across all 17 async test files with no missing decorators, and coroutine-never-awaited promoted to a hard error with a clean run.

docs/reviews/2026-09-02-code-review.md

@@ -0,0 +1,896 @@
+# Architecture & Code Review Report
+
+**Repository Target:** `C:\Github\transcription\`
+**Target Stack:** Python 3.12+ | FastAPI | NiceGUI | SQLModel/SQLAlchemy | Pydantic V2 | asyncio | OpenRouter
+**Review Date:** 2026-09-02
+**Canonical Baseline:** V6.1 (`docs/index.md`)
+
+---
+
+## 0. Verification Commands and Outcomes
+
+All four commands were executed in this checkout before any finding was written. This report
+records the exact outcomes rather than assuming them.
+
+| Command | Outcome |
+| :--- | :--- |
+| `uv run pytest -q -m "not external"` | **410 passed, 0 failed, 0 errors** (exit 0) |
+| `uv run ruff check .` | **All checks passed!** |
+| `uv run ruff format --check .` | **191 files already formatted** |
+| `uv run ty check` | **All checks passed!** |
+
+The stated green baseline is real. No finding below is a test failure; every finding is a
+behavior, contract, or guard-coverage defect that the passing suite does not detect.
+
+---
+
+## 1. Executive Summary
+
+- **The system's core evidence guarantees hold.** `ExecutionAttempt` is genuinely append-only,
+  attempt numbering is allocated with bounded conflict retry, transport evidence is captured at
+  the HTTP boundary before SDK parsing, and header persistence uses a true allowlist. Provenance
+  invariant families A–E and G pass.
+- **Both competing atomicity invariants in `services/workflows.py` are real and both guards
+  genuinely enforce them.** I injected-fault-verified the tests rather than trusting the
+  docstrings: `test_pipeline_atomicity.py` fails on a split final-page commit, and
+  `test_workflows_reliability.py:318` reads intermediate attempts through a *separate session*,
+  so it would fail if intermediate pages stopped committing individually.
+- **The most significant defect is a privacy leak that a prior review believed it had closed.**
+  The 2026-08-23 review moved root-cause text out of `AppError.message` into `AppError.detail`
+  to keep filesystem paths away from users. That text now reaches users anyway, because the UI
+  renders `ExecutionAttempt.error_detail` verbatim (HIGH-01). The leak was relocated, not closed.
+- **A second, independent path leak exists in five explicit `raise` sites** that the existing
+  guard never covered — it tests only `classify_unexpected_error` (HIGH-02).
+- **Provenance invariant family F (path safety) fails**, and it fails *inconsistently within one
+  file*: `sources_page.py:443` carefully sanitizes a stored path through
+  `public_media_path_label`, then `sources_page.py:484` dumps raw `error_detail` forty lines later.
+- **The orphan sweep does not do what its docstring claims.** It matches definitions by bare name,
+  so an entirely dead *module* passes whenever its function names collide with live ones.
+  `ui/pages/tags_page.py` is the proof: 93 lines never imported by anything (MED-01/LOW-01).
+- **On the three flagged open items:** the V4/V6.1 doc drift is confirmed (MED-02); the
+  `.env.production` coupling is real but currently correct and loud-failing, so Medium not High
+  (MED-03); and the Tags roadmap is **right** — the route is genuinely not registered, so the
+  module is dead code rather than a live retired route.
+- **Two latent concurrency defects carry ordering constraints** and must be fixed *before* the
+  changes that would make them live (MED-04, MED-05), not after.
+- **Guidance-file accuracy:** the recently revised `.github/instructions/*` files were verified
+  against code rather than trusted. They are accurate as written; the code is what diverges from
+  them. The one exception is that `error-handling.instructions.md` states a `detail` rule the UI
+  layer has never followed, which makes it an unenforced claim rather than a wrong one.
+
+---
+
+## 2. Executive Architecture Assessment
+
+**Verdict: architecturally sound, with a concentrated failure in the *last mile* of error
+presentation.**
+
+Domain cohesion and dependency direction are good and, unusually, mechanically enforced.
+`test_service_boundaries.py` and `test_ui_boundaries.py` AST-scan for violations using
+*allowlists* rather than blocklists, which is the correct choice — a newly added persistence
+helper cannot slip through under an unlisted name. `workflows.py` imports only the abstract
+`providers` types and never `openrouter`, so provider details genuinely stop at the adapter.
+Transaction ownership is explicit and well-reasoned: `ServiceBase._finalize` commits for
+service-owned sessions and flushes for caller-owned ones, which is what lets orchestration
+modules compose multi-aggregate writes without services importing each other.
+
+The evidence layer is the strongest part of the system and shows real care. The distinction
+between transport response, SDK-parsed response, and normalized metadata is maintained in code,
+not just in prose — `_CapturingAsyncClient` exists specifically to retain the exact wire body
+before the SDK can discard unknown fields, and `TransportEvidence(response_received=False)`
+explicitly represents "no response was received" rather than conflating it with an empty one.
+
+The weakness is at the boundary where internal diagnostic text becomes pixels. Every layer
+*below* the UI respects the message/detail split; the UI layer reads the internal field directly
+and renders it. The architecture defines the contract correctly and then has no enforcement at
+the one layer that violates it.
+
+**Top systemic risks:**
+
+1. **Internal diagnostic text reaches users through the evidence display path** (HIGH-01). The
+   rule is documented in three places and enforced in none of them at the UI boundary.
+2. **Path-safety discipline is applied per-call-site rather than structurally** (HIGH-02, HIGH-01).
+   It is correct wherever someone remembered; there is no guard that makes forgetting fail.
+3. **Guard coverage is narrower than guard docstrings claim.** Two guards
+   (`test_orphan_sweep.py`, `test_errors.py`) assert something meaningfully weaker than the
+   invariant they are named for, which converts them into a false sense of enforcement.
+4. **Worker safety currently rests on single-process sequential execution, not on configuration**
+   (MED-04, MED-05). Nothing is wrong today; two plausible future changes each make something wrong.
+
+---
+
+## 3. Findings by Severity
+
+### Critical Severity
+
+*None.* No evidence loss, append-only violation, secret leakage, or silent-wrong-output defect
+was found. The candidates in this class (provider evidence mis-attribution, stale-job double
+processing) are latent and are reported at High/Medium with their unblocking conditions.
+
+---
+
+### High Severity
+
+#### [HIGH-01] Internal-only `error_detail` is rendered directly to users, reopening the leak the 2026-08-23 fix was meant to close
+
+- **Location:**
+  - Write side: `src/transcription/errors.py:99-139` (`classify_unexpected_error` → `detail`, `format_error_detail` → persisted text)
+  - Persist: `src/transcription/services/workflows.py:720` (`error_detail=format_error_detail(page.error)`)
+  - **Render (Source Detail):** `src/transcription/ui/pages/sources_page.py:481-484`
+  - **Render (Sources list):** `src/transcription/ui/pages/sources_page.py:121` → `src/transcription/ui/components/table/sources.py:39,90-95` ("Error Detail" column)
+  - **Render (Maintenance):** `src/transcription/ui/pages/settings_page.py:562`, written by `src/transcription/services/maintenance.py:206`
+  - Contract violated: `docs/error_handling.md:107-114`; `.github/instructions/error-handling.instructions.md:86`; `docs/invariant/error_handling.md:59`; `docs/invariant/ai_evidence_and_provenance.md:103`
+
+- **Reachability:** **Live.** Concrete path, no configuration required: a page fails with any
+  non-`AppError` exception → `workflows.py:388` calls `classify_unexpected_error(exc)` →
+  `errors.py:118` sets `detail=f"{type(exc).__name__}: {exc}"` → `format_error_detail`
+  (`errors.py:135-139`) emits `... | detail=OSError: [Errno 13] Permission denied: '/app/uploads/documents/<uuid>/page-1.jpg' | ...`
+  → persisted to `ExecutionAttempt.error_detail` → rendered verbatim at
+  `sources_page.py:484` and in the `/sources` table column. A SQLAlchemy `OperationalError`
+  carries the database path by the same route.
+
+- **Problem & Consequence:** `docs/error_handling.md:110` states `detail` is *"Internal only"* and
+  that its only surfaces are `format_error_detail` (evidence) and logs;
+  `error-handling.instructions.md:86` says *"Never rendered to users or serialized into an
+  envelope."* The UI reads it anyway. The consequence is not hypothetical drift — it is the
+  precise defect the previous review's fix existed to prevent. That fix made `message` generic and
+  moved the root cause to `detail` on the stated grounds that `detail` never reaches users. That
+  premise was never true: `error_detail` had a UI consumer the whole time. The result is that the
+  filesystem-path leak was relocated from the notification banner to the Source Detail card and
+  the Sources table, while the test suite records the leak as fixed
+  (`tests/test_errors.py:56-78`).
+
+  The inconsistency is visible inside a single file: `sources_page.py:443` deliberately routes a
+  stored path through `public_media_path_label` (`ui/components/media_urls.py:58-72`), which
+  correctly degrades an absolute path to its bare filename — and then `sources_page.py:484`
+  renders unsanitized text that may contain an absolute path.
+
+- **Blast Radius:** Enumerated by grepping every reader of `.detail` and `error_detail`:
+  - `errors.py:137` — `format_error_detail`, the only reader of `AppError.detail`. **Must keep the root cause.**
+  - `services/workflows.py:720` — the only writer of `ExecutionAttempt.error_detail`.
+  - `services/maintenance.py:206` — the only writer of `MaintenanceRun.error_detail`.
+  - `services/evidence.py:195` — `build_evidence_export` emits `error_detail`. Export is an
+    operator-initiated evidence artifact; per invariant 3.7.1 it **must** retain it.
+  - `db/models.py:508-522` — `Source.latest_error_detail` projection, consumed only by `sources_page.py:121`.
+  - `ui/pages/sources_page.py:481-484`, `ui/components/table/sources.py`, `ui/pages/settings_page.py:562` — the three render sites.
+  - Tests asserting on persisted text: `tests/test_v42_evidence.py:284`,
+    `tests/services/test_workflows_reliability.py` (timeout detail),
+    `tests/services/test_maintenance_service.py`. A fix that changes *what is stored* breaks these;
+    a fix that changes *what is displayed* does not.
+
+- **Recommendation — two invariants conflict here; both must be named.**
+
+  **Invariant 1 (evidence):** `ExecutionAttempt.error_detail` must retain the root cause.
+  `docs/requirements.md:30` (REQ-4-021) and `docs/invariant/ai_evidence_and_provenance.md:33`
+  require it; guarded by `tests/test_v42_evidence.py::test_attempts_are_append_only_and_exported_with_integrity`
+  and `tests/services/test_workflows_reliability.py`.
+
+  **Invariant 2 (privacy):** user-facing surfaces must not expose local filesystem details.
+  `docs/invariant/error_handling.md:59`; guarded (partially) by
+  `tests/test_errors.py::test_unexpected_error_does_not_leak_filesystem_paths`.
+
+  **The over-correction to avoid is stripping root-cause text out of `detail` or
+  `format_error_detail` to make the UI safe.** That is exactly the mistake documented in the
+  reviewer skill's worked example, and it would silently destroy the provenance record this
+  system exists to preserve while making every guard still pass.
+
+  Fix at the **render** boundary, not the write boundary. Add a presentation-layer projection and
+  route all three UI sites through it, leaving the persisted evidence untouched:
+
+  ```python
+  # src/transcription/ui/components/error_presenter.py  (new)
+  def display_failure_detail(error_detail: str | None) -> str | None:
+      """Render persisted failure detail without machine-local paths.
+
+      `ExecutionAttempt.error_detail` is provenance and keeps the full root cause
+      (docs/error_handling.md). This projection is the only thing a page may show.
+      """
+  ```
+
+  It should preserve the `[category]`, `suggestion=`, and `error_id=` segments (which are what
+  make the display actionable) and reduce any absolute path inside `detail=` to its basename,
+  mirroring `public_media_path_label`. The operator keeps diagnosability — required by
+  `docs/ui/pages/sources.md:43` and `docs/requirements.md:59` (REQ-6-014) — without the container
+  filesystem layout being published to the browser.
+
+  Then decide and record which resolution was chosen: either the UI shows the sanitized
+  projection (recommended), or `docs/error_handling.md:107-114` and
+  `error-handling.instructions.md:86` are revised to state that operator-facing evidence displays
+  may render `error_detail` **and** that the guarantee moves to "no machine-local detail ever
+  enters `detail`" — which would be a much harder guarantee to keep. Do not leave the current
+  state, where the docs claim one thing and three pages do another.
+
+- **Effort:** M
+
+---
+
+#### [HIGH-02] Absolute filesystem paths are embedded in user-facing `AppError.message` at five explicit raise sites
+
+- **Location:**
+  - `src/transcription/services/sources.py:856` — `f"Prompt file not found: {prompt_path}"`
+  - `src/transcription/services/sources.py:864` — `f"Prompt file is empty: {prompt_path}"`
+  - `src/transcription/services/sources.py:914` — `f"Source file not found: {path}"`
+  - `src/transcription/services/prompts.py:99` — `f"Prompt directory is unavailable: {root}"`
+  - `src/transcription/services/prompts.py:186-191` — `_filesystem_error` builds `f"{message}: {exc}"`
+  - Contract violated: `.github/instructions/error-handling.instructions.md:74,85`; `docs/invariant/error_handling.md:59`
+
+- **Reachability:** **Live**, on an ordinary user path. `sources.py:845` resolves
+  `prompt_root = runtime_settings.prompt_dir.resolve()`, so `prompt_path` is absolute
+  (`/app/prompts/transcribe_document.md` in the container). `load_prompt_text` is invoked by
+  `build_prompt_execution` (`sources.py:829-831`), which runs on **every document upload** via
+  `services/store.py:94` and `store.py:162`. The resulting `PromptLoadError` is an `AppError`
+  subclass, so it flows through `run_ui_action` → `show_error`
+  (`ui/components/error_presenter.py:51-66`), which renders `error.message` into both a
+  `ui.notify` banner and a card label, and through `build_error_envelope` (`errors.py:88-96`)
+  into API responses.
+
+- **Problem & Consequence:** `error-handling.instructions.md:85` requires `message` to *"Stay
+  generic. Never embed exception text, provider payloads, or filesystem paths."* These five sites
+  embed exactly that. `prompts.py:186-191` violates the rule in **both** directions at once: it
+  puts `{exc}` — an `OSError` whose `str()` includes the offending filename — into `message`, and
+  it sets **no `detail=`**, so the internal field that is supposed to carry the root cause is
+  empty while the user-facing field carries all of it.
+
+  This is not a new regression; it is coverage that the existing guard never had.
+  `tests/test_errors.py:56-78` verifies only that `classify_unexpected_error` — the *catch-all*
+  path — does not leak. Every deliberate `raise SomeError(f"... {path}")` in the codebase is
+  outside its scope, so the suite reports the invariant as enforced while five live sites violate it.
+
+- **Blast Radius:** Verified by grepping all consumers of these exception types.
+  `PromptLoadError`/`PromptStoreError`/`TranscriptionError` messages are consumed by:
+  `ui/components/error_presenter.py:55,63` (render), `errors.py:92` (API envelope),
+  `errors.py:135` (`format_error_detail` → evidence). Because the recommended change *adds* a
+  `detail` and *shortens* `message`, `format_error_detail` output still contains the path — so
+  evidence value is preserved, not reduced. Tests asserting on these messages:
+  `tests/test_prompts.py`, `tests/services/test_prompt_store.py`,
+  `tests/services/test_transcription_service.py`. These assert on message prefixes
+  (`"Prompt file not found"`), not on the interpolated path, and were checked to survive the change —
+  but re-run them, since `prompts.py:186` currently produces a message whose suffix some
+  assertion could depend on.
+
+- **Recommendation:** Apply the pattern `errors.py:113-119` already establishes — generic
+  `message`, root cause on `detail`, `raise ... from exc`. Use `path.name` when a filename is
+  genuinely useful to the user.
+
+  ```python
+  # sources.py:855 — before
+  raise PromptLoadError(f"Prompt file not found: {prompt_path}", ...)
+  # after
+  raise PromptLoadError(
+      f"Prompt file not found: {prompt_path.name}",
+      category=ErrorCategory.INFRA_PERSISTENT,
+      suggestion="Verify PROMPT_DIR and prompt file configuration, then retry.",
+      detail=f"Prompt file missing at {prompt_path}",
+  )
+
+  # prompts.py:186 — before
+  return PromptStoreError(f"{message}: {exc}", category=..., suggestion=...)
+  # after
+  return PromptStoreError(
+      message,
+      category=ErrorCategory.INFRA_PERSISTENT,
+      suggestion="Check prompt directory permissions and available disk space, then retry.",
+      detail=f"{type(exc).__name__}: {exc}",
+  )
+  ```
+
+  Then widen the guard so this class cannot recur — see MED-07. Note the dependency: HIGH-02 and
+  HIGH-01 must be fixed **together**, because moving the path from `message` to `detail` while the
+  UI still renders `error_detail` relocates the leak instead of closing it. That is the same
+  mistake that produced HIGH-01.
+
+- **Effort:** S (fix) / M (with the guard)
+
+---
+
+### Medium Severity
+
+#### [MED-01] The orphan sweep matches by bare name and therefore cannot detect a dead module
+
+- **Location:** `tests/test_orphan_sweep.py:101-163` (`_public_definitions`, `_orphans`)
+- **Reachability:** **Live** — the guard is running now and reporting a clean sweep that is not clean.
+- **Problem & Consequence:** `_public_definitions()` keys definitions by bare name
+  (`definitions[node.name]`, line 111) and `_orphans()` marks a definition referenced if that
+  bare name appears **anywhere** in `src/`, `tests/`, or `tools/` (lines 157-162). Two different
+  modules that define the same public name are therefore indistinguishable, and neither can ever
+  be reported as an orphan.
+
+  `src/transcription/ui/pages/tags_page.py` demonstrates the consequence. Its only public
+  definition is `register_page` (line 18). Seven live page modules define a function of the same
+  name and `ui/__init__.py:37-43` calls all seven — so `register_page` is heavily referenced and
+  `tags_page.register_page` is scored as reachable. In fact **nothing imports `tags_page` at all**
+  (verified: the only repo-wide references to the module are the file itself and
+  `tests/ui/test_tags_page.py`, which merely asserts the route 404s). 93 lines of code, including a
+  lazy-load-unsafe relationship traversal at `tags_page.py:71-74`, sit outside the sweep's reach.
+
+  The sweep also never asks whether a *module* is imported, only whether its definitions' names
+  appear somewhere — so this is a structural gap, not a one-off miss.
+
+- **Blast Radius:** `tests/test_orphan_sweep.py` only; `KNOWN_ORPHANS` entries are keyed by the
+  same bare/dotted names and would need re-keying if qualification is added. Expect the stricter
+  sweep to surface additional true orphans on first run — triage them into `KNOWN_ORPHANS` with
+  rationales rather than weakening the check.
+- **Recommendation:** Qualify definitions by module (`f"{module_path}:{name}"`) and add a separate,
+  cheap module-reachability pass: a module under `src/transcription/` is reachable if any other
+  module imports it, or it is a declared entrypoint (`app.py`, `__main__.py`, `worker_service.py`).
+  Report unreachable modules as orphans in their own right. Also fix
+  `test_public_definitions_are_discovered` (line 169), whose `>= 420` snapshot threshold is a
+  weak assertion that drifts upward silently — the 2026-08-23 review already flagged the same
+  pattern at the then-current `>= 200` and it was raised rather than replaced.
+- **Effort:** M
+
+---
+
+#### [MED-02] Canonical invariant document declares a V4 baseline while the canonical baseline is V6.1
+
+- **Location:** `docs/invariant/ai_evidence_and_provenance.md:130`
+- **Reachability:** **Live** (documentation), no runtime impact.
+- **Problem & Consequence:** Section 6.1 reads *"Canonical V4 architecture, schema, requirements,
+  and error-policy documents define how current behavior satisfies this invariant."*
+  `docs/index.md:1,29-32` establishes V6.1 as the baseline and states that every canonical document
+  asserts the same baseline. This is the **ownership clause of the invariant that governs the
+  entire evidence model** — the clause that tells a reader which documents are authoritative — and
+  it points at a superseded generation. A reader following it lands on stale authority precisely
+  when resolving an evidence question, which is the highest-stakes case.
+
+  A baseline-currency guard **does** exist —
+  `tests/test_meta_contract_guards.py::test_canonical_docs_declare_one_consistent_baseline`
+  (lines 89-112) — and `docs/invariant/ai_evidence_and_provenance.md` is **not** in
+  `BASELINE_SCAN_EXCLUSIONS` (lines 56-64), so the file is scanned. The claim escapes for two
+  independent reasons, either of which alone would be sufficient:
+  1. `_CURRENT_VERSION_CLAIM` (line 67) matches only the words `current` or `active` before a
+     version. This line says "**Canonical** V4", a third phrasing the pattern does not know.
+  2. Both patterns require `V(\d+\.\d+)` — a mandatory minor version. The bare token `V4` cannot
+     match either regex under any phrasing.
+
+  The guard is therefore not absent but *phrase-shaped*: it enforces currency only for the two
+  sentence forms someone thought of, against version strings that carry a minor. That is a weaker
+  property than its docstring implies ("Every canonical doc that names the current baseline must
+  name the same one").
+- **Blast Radius:** Documentation only; no code reads this string. Widening the guard's patterns
+  will re-scan all canonical docs — expect it to surface further stale mentions on first run
+  (`docs/architecture.md`, `docs/schema.md`, `docs/requirements.md`, and `docs/error_handling.md`
+  each contain 2-3 version tokens), which should be triaged rather than excluded.
+- **Recommendation:** Two parts, and the second matters more than the first.
+  1. Change "Canonical V4" to "Canonical V6.1" at line 130.
+  2. Fix the guard's shape rather than adding a third phrase to the list. Accept an optional minor
+     (`V(\d+)(?:\.(\d+))?`) and invert the matching: flag **every** `V<n>` token in a scanned
+     canonical doc that is not the declared baseline, rather than only those preceded by an
+     approved adjective. Phrase-list matching fails open — each new phrasing silently reopens the
+     hole — whereas token matching fails closed and forces an explicit exclusion.
+
+  See §8.1 for the alternative the maintainer is considering: dropping version labels from
+  canonical docs entirely, which removes the failure mode instead of guarding it.
+- **Effort:** S
+
+---
+
+#### [MED-03] Settings resolve `.env.production` relative to the process working directory, and the isolation fix exists only in the test harness
+
+- **Location:** `src/transcription/config.py:66-75` (`env_file=".env.production"`);
+  workaround at `tests/conftest.py:27-50`; guarded by `tests/test_config_isolation.py`;
+  depended on by `.github/workflows/quality-gate.yml` and `docker-compose.production.yml`
+- **Reachability:** **Live but currently correct.** I verified the production path rather than
+  assuming it: `Dockerfile` sets `WORKDIR /app` in the runtime stage, and
+  `docker-compose.production.yml` mounts `./.env.production` to `/app/.env.production` for both the
+  `app` and `worker` services, so the relative path resolves correctly today.
+- **Problem & Consequence:** Correct configuration loading depends on an **implicit, undocumented
+  contract between `config.py` and the process working directory.** Nothing in `config.py` states
+  it, and nothing tests it. The failure mode is not silent — `openrouter_api_key` is required with
+  no default, so a wrong cwd produces a `ValidationError` at startup rather than a partially
+  configured process — which is why this is Medium rather than High.
+
+  The more telling symptom is what the coupling forced on the test harness. `conftest.py:45-50`
+  cannot escape it by passing an argument; it must **mutate the Pydantic class-level
+  `model_config` dict at runtime** and restore it in a `finally`. That is a global, order-sensitive
+  side effect adopted because the module offers no seam. It also silently repairs a second
+  consumer: `ui/runtime_settings_store.py:402` reads the same `Settings.model_config["env_file"]`
+  to decide where the Settings page writes. Two subsystems are coupled through a mutable class
+  attribute.
+- **Blast Radius:** Every `Settings` construction. Consumers of `model_config["env_file"]`:
+  `ui/runtime_settings_store.py:402` (write-target resolution, contract documented at
+  `docs/ui/pages/settings.md:27`) and `tests/conftest.py:45-50`. A change must preserve the
+  documented three-step resolution order — explicit override, `RUNTIME_SETTINGS_ENV_FILE`, then the
+  configured default — or `docs/ui/pages/settings.md:27` becomes wrong.
+- **Recommendation:** Introduce one explicit resolution function that both `Settings` construction
+  and `runtime_settings_store` call, honoring an `ENV_FILE` environment variable and falling back
+  to a path anchored to a known root rather than to `os.getcwd()`. Tests then pass a path instead of
+  mutating class state, and `tests/test_config_isolation.py` can assert against the seam rather
+  than against the monkeypatch. If instead the cwd contract is accepted as deliberate, document it
+  in `config.py` and in `docs/production-runbook.md` and add a guard asserting `WORKDIR`/cwd
+  alignment — an implicit contract with a container image is exactly the kind of rule the invariant
+  routing table exists to place.
+- **Effort:** M
+
+---
+
+#### [MED-04] Stale-job reclaim threshold is not derived from maximum job duration; safety currently comes from single-process sequencing
+
+- **Location:** `src/transcription/config.py:116-117`
+  (`worker_provider_timeout_seconds=30.0`, `worker_stale_job_seconds=30.0`);
+  sweep at `src/transcription/worker.py:222-228`; reclaim at
+  `src/transcription/services/jobs.py:242-268`
+- **Reachability:** **Latent.** Unblocked by *either* of: (a) running more than one worker replica
+  (adding `deploy.replicas > 1` to the `worker` service in `docker-compose.production.yml`), or
+  (b) setting `RUN_EMBEDDED_WORKER=true` on the `app` service while the standalone `worker`
+  container is also running. It is safe today only because
+  `docker-compose.production.yml` sets `RUN_EMBEDDED_WORKER: "false"` on `app` and defines exactly
+  one `worker`, and because within a single loop `run_worker_loop` awaits
+  `process_next_queued_job` to completion before returning to the stale sweep — so the sweep can
+  never observe a job that this same process is actively working.
+- **Problem & Consequence:** The stale threshold (30s) **equals** the per-page provider timeout
+  (30s), leaving zero margin even for a single-page job. A multi-page document is legitimately
+  `PROCESSING` for up to N × 30s. `Job.date_updated` carries an `onupdate`
+  (`db/models.py:372-375`), but between the initial claim and the terminal write the only touch
+  is `sources.py:522-523` reassigning `job.provider`/`job.model` to values they usually already
+  hold, which SQLAlchemy resolves to no net change and therefore no `UPDATE`. I did not empirically
+  confirm the no-`UPDATE` behavior, so treat that specific step as unverified — but the finding does
+  not depend on it, because even a per-page refresh leaves only a 30s margin against a 30s timeout.
+
+  With a second concurrent worker, the sweep would requeue a job that is mid-provider-call. Both
+  workers then process the same job, producing duplicate `ExecutionAttempt` rows for the same
+  logical work and racing terminal status writes. Append-only history would be *preserved* but no
+  longer *faithful*: the evidence would show attempts that do not correspond to distinct
+  application decisions.
+
+  This is worth flagging because `jobs.py:191-197` explicitly implements and documents
+  `SKIP LOCKED` row locking "so concurrent workers never contend for the same job." The claim path
+  is built for multi-worker operation; the reclaim path is not. A reader who trusts the claim
+  docstring would reasonably scale the worker.
+- **Blast Radius:** `requeue_stale_processing_jobs` has one production caller (`worker.py:226`) and
+  tests in `tests/test_worker.py` and `tests/services/test_job_service.py`. Changing the *default*
+  affects `tests/test_config.py` declared-defaults assertions — check those before editing the default.
+- **Recommendation:** **Fix before adding a second worker replica, not after.** Two parts:
+  (1) Make the threshold a function of the real bound rather than a coincidental peer of the
+  page timeout — at minimum default `worker_stale_job_seconds` to a multiple of
+  `worker_provider_timeout_seconds` with headroom, and add a model validator rejecting a stale
+  threshold at or below the provider timeout.
+  (2) Preferably make reclaim heartbeat-based: have `_persist_page_outcome` bump `Job.date_updated`
+  explicitly so liveness reflects progress rather than elapsed time since claim.
+  Add a guard asserting a multi-page job in flight is not reclaimed by a concurrently-invoked sweep.
+- **Effort:** M
+
+---
+
+#### [MED-05] Provider evidence capture is per-instance mutable state, making the adapter non-reentrant by contract
+
+- **Location:** `src/transcription/providers/openrouter.py:197-199, 264-267, 274-275, 297-298, 397-412`;
+  `_CapturingAsyncClient.last_response`/`last_body` at `openrouter.py:66-94`;
+  contract at `src/transcription/providers/base.py:110-118`
+  (`current_request_manifest`, `current_transport_evidence`)
+- **Reachability:** **Latent.** Unblocked by any concurrent `transcribe()` on a single adapter
+  instance — most plausibly by processing a job's pages in parallel (`workflows.py:274` is
+  currently a sequential `for` loop) or by any second consumer sharing one
+  `SourceService.provider`. Verified safe today: `workflows.py:272` resolves one provider for the
+  loop and awaits each page; the worker's `ServiceBundle` (`worker.py:206`) is distinct from
+  `app.state.services` (`app.py:43`), so the UI cannot share the worker's adapter instance, and
+  the UI only enqueues jobs (`ui/pages/jobs_page.py:186-208`).
+- **Problem & Consequence:** The `TranscriptionProvider` protocol defines evidence retrieval as
+  "the most recent call" state read *after* the fact. `workflows.py:369-370` relies on this on the
+  timeout path, reading `provider.current_request_manifest` / `current_transport_evidence` when no
+  result object exists. Under concurrency, page B's response overwrites
+  `_CapturingAsyncClient.last_response` before page A's timeout handler reads it, and page A's
+  `ExecutionAttempt` is written with page B's transport evidence.
+
+  The consequence is **evidence mis-attribution** — a provenance-integrity failure, which this
+  project's own rubric treats as its most serious class. It would also be near-undetectable after
+  the fact: the attempt row would be well-formed, internally consistent, and wrong. The
+  application-level design that makes this safe (sequential pages) is not expressed in the
+  provider contract, so the constraint lives only in `workflows.py`'s loop structure.
+- **Blast Radius:** Changing the protocol touches `providers/base.py:102-136`,
+  `providers/openrouter.py:221-231`, the two read sites at `workflows.py:369-370`, and the fakes in
+  `tests/providers/test_openrouter.py`, `tests/services/test_workflows_reliability.py`, and
+  `tests/test_provider_boundaries.py`, all of which implement or assert the current property-based
+  contract.
+- **Recommendation:** **Fix before introducing any intra-job page concurrency.** The durable fix is
+  to stop returning evidence through instance state: attach `request_manifest` and
+  `transport_evidence` to the raised exception on every failure path — which `ProviderError`
+  already supports (`providers/base.py:18-29`) and which the timeout path cannot currently use
+  because `asyncio.wait_for` raises `TimeoutError` from outside the adapter. A narrower option is
+  to have `transcribe()` accept a caller-owned capture sink so evidence is scoped to the call
+  rather than to the adapter. As an immediate, near-zero-cost step, document the non-reentrancy on
+  the protocol in `providers/base.py` so the constraint is visible where it is depended upon.
+- **Effort:** M
+
+---
+
+#### [MED-06] Provider error bodies reach user-facing text while three provider failure paths persist no `detail`
+
+- **Location:** `src/transcription/services/sources.py:923-947` (`handle_transcription_errors`);
+  message construction at `src/transcription/providers/openrouter.py:414-431`
+  (`_transport_error_message`)
+- **Reachability:** **Live** for the message half (any provider failure during a UI-initiated
+  transcription surfaces through `show_error`).
+- **Problem & Consequence:** Two mirrored halves of the same rule are broken in one function.
+  - `sources.py:943` builds `f"Provider transcription failed: {exc}"`, and `exc` is a
+    `ProviderError` whose message may embed up to 500 characters of the provider's error body
+    (`openrouter.py:430`). That is a provider payload in `message`, which
+    `error-handling.instructions.md:85` explicitly forbids.
+  - None of the three handlers (lines 929, 935, 942) passes `detail=`. Per
+    `error-handling.instructions.md:89-92`, omitting it degrades the provenance record.
+
+  I checked whether the provenance half is actually harmful before reporting it, and it is
+  **substantially mitigated**: `workflows.py:391` calls `_find_provider_error`, which walks
+  `__cause__`/`__context__` (`workflows.py:806-813`) to recover the original `ProviderError` and
+  persists its `transport_evidence` — status code, safe headers, and the exact response body — onto
+  the attempt. So the root cause is preserved in transport evidence even though `error_detail` is
+  thin. This is why the finding is Medium rather than High. The residual cost is that the
+  human-readable failure summary is uninformative for the two paths (`ProviderAuthError`,
+  `ProviderResponseError`) whose messages are entirely generic.
+- **Blast Radius:** `handle_transcription_errors` is used on the transcription path in
+  `sources.py`; `TranscriptionError.message` is consumed by `error_presenter.show_error`,
+  `build_error_envelope`, and `format_error_detail`. Assertions on these messages live in
+  `tests/services/test_transcription_service.py` and `tests/providers/test_openrouter.py`.
+- **Recommendation:** Move the interpolated provider text from `message` to `detail` on all three
+  handlers, keeping the generic message the other two already use:
+  ```python
+  except ProviderError as exc:
+      raise TranscriptionError(
+          "Provider transcription failed",
+          category=ErrorCategory.EXTERNAL_PROVIDER,
+          suggestion="Retry the transcription from jobs. If repeated, check provider availability.",
+          retriable=True,
+          detail=f"{type(exc).__name__}: {exc}",
+      ) from exc
+  ```
+  Apply the same `detail=` addition to the `ProviderAuthError` and `ProviderResponseError`
+  handlers. Note the interaction with HIGH-01: until the render boundary is sanitized, moving text
+  into `detail` still reaches users through the `error_detail` display. Sequence accordingly.
+- **Effort:** S
+
+---
+
+#### [MED-07] No deterministic guard covers the `message`/`detail` split at explicit raise sites
+
+- **Location:** `tests/test_errors.py:56-78`; rule at
+  `.github/instructions/error-handling.instructions.md:78-98`; canonical statement at
+  `docs/error_handling.md:102-115`
+- **Reachability:** **Live** — this coverage gap is what allowed HIGH-02 and MED-06 to exist in a
+  fully green suite.
+- **Problem & Consequence:** `docs/error_handling.md:115` names
+  `tests/test_errors.py::test_unexpected_error_does_not_leak_filesystem_paths` as the enforcement
+  for the message/detail split. That test exercises exactly one function,
+  `classify_unexpected_error`. Every direct `raise SomeAppError(...)` in `src/` — roughly 50 sites
+  by grep — is unenforced. The documentation therefore overstates the enforcement, which is worse
+  than having no guard: a contributor reading `error_handling.md:115` reasonably concludes the rule
+  is mechanically protected.
+
+  Per the reviewer skill, where a check is unenforced, recommending the deterministic test is
+  itself a finding.
+- **Blast Radius:** Tests only.
+- **Recommendation:** Add an AST guard, `tests/test_error_message_safety.py`, that scans `src/`
+  for `raise <AppError subclass>(...)` and fails when the first positional argument is an f-string
+  containing a formatted value whose name matches a path-like or exception-like identifier
+  (`path`, `_path`, `root`, `dir`, `exc`, `err`, `e`). Model it on the existing AST guards, which
+  are the established pattern here (`test_ui_boundaries.py`, `test_service_boundaries.py`,
+  `test_orphan_sweep.py`). Pair it with a second guard asserting that no UI module reads
+  `error_detail` without routing through the sanitizing projection from HIGH-01 — that one closes
+  the render side, which is where the real leak is.
+- **Effort:** M
+
+---
+
+### Low Severity
+
+#### [LOW-01] `ui/pages/tags_page.py` is dead code; the V6.1 roadmap is correct
+
+- **Location:** `src/transcription/ui/pages/tags_page.py` (93 lines);
+  registration list at `src/transcription/ui/__init__.py:37-43`
+- **Reachability:** **Not reachable.** This resolves the flagged open item: the route is genuinely
+  **not** registered. `register_pages` calls seven page registrars and `tags_page` is not among
+  them; nothing anywhere imports the module. `docs/roadmap_plan.md:47` ("Retire the Tags page") is
+  accurate, and `tests/ui/test_tags_page.py` correctly asserts `/ui/tags` returns 404 — though it
+  passes trivially, since an unimported module cannot register anything.
+- **Problem & Consequence:** No runtime risk; purely stranded code. It is worth noting that if it
+  *were* ever re-registered, `tags_page.py:71-74` traverses `document.document_tags` and
+  `link.tag_ref` inside a page render, and those relationships are configured `lazy="raise"`
+  (`docs/architecture.md:200-203`) — so re-enabling this module without adding eager loads to
+  `list_documents` would raise on first render.
+- **Recommendation:** Delete `src/transcription/ui/pages/tags_page.py`. Retain
+  `tests/ui/test_tags_page.py` as the retirement guard. Fixing MED-01 first would make this
+  finding reproducible by the suite rather than by manual inspection.
+- **Effort:** S
+
+#### [LOW-02] `benchmarking.py` ships in the runtime package but is referenced only by tests
+
+- **Location:** `src/transcription/benchmarking.py` (69 lines); sole consumers
+  `tests/test_v42_evidence.py:15-16` (`EditorialAssessment`, `score_transcription`)
+- **Reachability:** Live as importable API; never invoked by application code.
+- **Problem & Consequence:** No defect. It supports the model-evaluation policy in
+  `docs/invariant/ai_evidence_and_provenance.md:113-126`, which is legitimate, but it currently has
+  no production caller and no tooling entrypoint, so it is indistinguishable from drift.
+- **Recommendation:** Either move it under `tools/` alongside the other operator utilities, or add
+  a `KNOWN_ORPHANS`-style rationale recording that it is retained as the evaluation-policy
+  implementation. Do not silently keep it unlabeled.
+- **Effort:** S
+
+#### [LOW-03] Two overlapping prompt error types split across modules
+
+- **Location:** `src/transcription/services/errors.py:15-16` (`PromptLoadError`) and
+  `src/transcription/services/prompts.py:19` (`PromptStoreError`)
+- **Reachability:** Live; no misbehavior observed.
+- **Problem & Consequence:** `services/errors.py:1-8` documents itself as the neutral home for
+  exceptions raised by more than one service, precisely so a caller's `except` clause does not
+  change when an operation moves. `PromptStoreError` is defined outside that module and covers an
+  overlapping domain (prompt file access), so a caller wanting to handle "any prompt failure" must
+  import from two modules and know which is which. `sources.py:855` raises `PromptLoadError` for a
+  missing prompt file while `prompts.py:132` raises `PromptStoreError` for the same condition
+  reached through the Settings page.
+- **Recommendation:** Move `PromptStoreError` into `services/errors.py` next to `PromptLoadError`,
+  or make one a subclass of the other so a single `except` covers prompt failures. Low urgency; do
+  it opportunistically when HIGH-02 touches both files anyway.
+- **Effort:** S
+
+---
+
+## 4. Architectural Drift & Gap Analysis
+
+| Area / Component | Direction | Documented / Intended Rule | Actual Implementation State | Severity | Recommended Resolution |
+| :--- | :--- | :--- | :--- | :--- | :--- |
+| Error presentation | `doc->code` | `docs/error_handling.md:110` — `detail` is internal only, surfaced by `format_error_detail` and logs | `sources_page.py:484`, `table/sources.py:90`, `settings_page.py:562` render `error_detail` verbatim to users | High | Sanitizing render projection (HIGH-01); do **not** strip `detail` |
+| User-facing messages | `doc->code` | `invariant/error_handling.md:59` — no local filesystem detail in user-facing messages | 5 live sites interpolate absolute paths into `AppError.message` | High | Generic `message`, path on `detail` (HIGH-02) |
+| Evidence invariant ownership | `doc->doc` | `docs/index.md:1` — baseline is V6.1 | `invariant/ai_evidence_and_provenance.md:130` names "Canonical V4"; the currency guard scans the file but its regexes match neither the phrasing nor a minor-less `V4` | Medium | Update text; make the guard token-based, or drop version labels entirely (MED-02, §8.1) |
+| Enforcement claim | `doc->code` | `docs/error_handling.md:115` — split "Enforced by `tests/test_errors.py::…`" | That test covers only `classify_unexpected_error`; explicit raises unguarded | Medium | Add AST guard (MED-07) |
+| Orphan sweep | `doc->code` | `test_orphan_sweep.py:1-13` — sweep is "deterministic" and "conservative" | Bare-name matching; cannot see a dead module (`tags_page.py`) | Medium | Qualify by module + module-reachability pass (MED-01) |
+| Worker scaling | `code->doc` | `jobs.py:191-197` — `SKIP LOCKED` so "concurrent workers never contend" | Claim path is multi-worker-safe; stale-reclaim path is not | Medium | Derive stale threshold from job duration; document single-worker constraint until fixed (MED-04) |
+| Provider adapter contract | `code->doc` | `providers/base.py:110-118` — evidence read as "most recent call" state | Contract is silently non-reentrant; safety lives in `workflows.py`'s sequential loop | Medium | Scope evidence to the call; document non-reentrancy (MED-05) |
+| Settings env file | `code->doc` | `config.py:66-75` — `env_file=".env.production"` | Correctness depends on an undocumented cwd contract with `Dockerfile` `WORKDIR /app` | Medium | Explicit resolver seam, or document + guard the contract (MED-03) |
+| Tags page | *(no drift)* | `roadmap_plan.md:47` — Tags page retired | Route genuinely unregistered; module is stranded code | Low | Delete the module (LOW-01) |
+
+---
+
+## 5. Invariant Inventory & Routing Recommendations
+
+| Invariant / Constraint | Current Location | Recommended Target Layer | Rationale |
+| :--- | :--- | :--- | :--- |
+| `detail`/`error_detail` never rendered to users | docs + instructions | **Deterministic test** + sanitizing projection | Stated in three documents and violated in three files; prose has demonstrably failed to hold it |
+| `message` carries no paths or exception text | instructions; partial test | **Deterministic test** (AST, all raise sites) | Existing guard covers one function; the gap produced HIGH-02 |
+| `ExecutionAttempt.error_detail` retains root cause | docs + `test_v42_evidence.py` | **Keep in tests** — already correct | Counterweight to the above; must be named in any fix so it is not over-corrected |
+| Intermediate pages commit individually | `workflows.py` docstring + `test_workflows_reliability.py:318` | **Keep in tests** — verified genuine | Cross-session read makes it a real durability assertion |
+| Final page atomic with terminal status | `services.instructions.md` + `test_pipeline_atomicity.py` | **Keep in tests** — verified genuine | Fault injection makes a split commit fail |
+| Canonical baseline version consistency | `docs/index.md` + `test_meta_contract_guards.py:89` | **Repair existing test, or remove the labels** | Guard exists but matches by approved phrase and requires a minor version, so it fails open on new phrasings (MED-02) |
+| Module-level reachability / dead modules | `test_orphan_sweep.py` (ineffective) | **Deterministic test** (repair existing) | Guard exists but cannot detect the case (MED-01) |
+| Stale threshold > max job duration | *(unenforced)* | **Config validator + test** | Currently a coincidence of two equal defaults (MED-04) |
+| Provider adapter non-reentrancy | *(unenforced, implicit)* | **Instructions** + protocol docstring | A design constraint callers must know before adding concurrency (MED-05) |
+| Env-file resolution independent of cwd | `tests/conftest.py` monkeypatch | **Code seam** + `docs/production-runbook.md` | A test-only fix for a production coupling is misrouted enforcement (MED-03) |
+
+---
+
+## 6. Stack-Specific Analysis
+
+**Python 3.12+.** Modern and consistent. PEP 695 generics are used correctly and non-trivially
+(`RegistryService[ModelT: RegistryEntry]` in `services/registry.py:58`, `UiActionOutcome[T]`,
+`_get_or_raise[ModelT]`), `type` statements appear in `db/session.py:15,50`, and `X | None` is
+used throughout. `structural Protocol` bounds (`RegistryEntry`, `WorkerNotifier`,
+`TranscriptionProvider`) are used to avoid type suppressions rather than to decorate. `ty` passes
+clean with no suppressions found. The two `# noqa` uses (`workflows.py:228` `PLR0915`,
+`workflows.py:383` `BLE001`) are both justified in context — the broad catch is a deliberate
+per-page containment boundary that immediately classifies and re-records.
+
+**FastAPI.** Lifespan is handled via `@asynccontextmanager` (`app.py:36`), not the deprecated
+`@app.on_event`. Session factories are injected through `Depends` (`SessionFactoryDep`,
+`db/session.py:50`) rather than reached as globals from routes. `api/errors.py` centralizes
+envelope translation. One residual: `get_settings` is `@cache`d and read as a module-level
+fallback in ~10 modules; this is acceptable given the documented restart-to-apply contract
+(`docs/ui/pages/settings.md:28`) but means the cache is process-lifetime and unclearable.
+
+**NiceGUI (pinned `3.13.0`).** The pin is a recorded release-stability decision and is not
+reported as a defect. Boundaries are enforced structurally: `test_ui_boundaries.py` uses an
+import **allowlist**, which is the right polarity. Blocking work is dispatched off the event loop
+via `run_blocking` (`settings_page.py:818,822`). The one boundary that is *not* enforced is
+presentation of internal fields (HIGH-01) — pages are prevented from touching persistence but not
+from rendering internal-only text.
+
+**SQLModel / SQLAlchemy.** Strong. `lazy="raise"` on relationships forces explicit eager loading;
+read paths declare `selectinload` chains with comments explaining *why* each is needed
+(`sources.py:309-316` is a good example). `expire_on_commit=False` (`db/session.py:28`) is set
+deliberately, which is what makes post-commit attribute access in `evidence.py:159-202` safe.
+`claim_next_queued_job` (`jobs.py:186-241`) branches correctly on dialect — `SKIP LOCKED` on
+PostgreSQL, conditional `UPDATE ... RETURNING` on SQLite — rather than assuming one engine.
+Attempt-number allocation uses `begin_nested()` with bounded retry (`sources.py:598-616`), the
+right pattern for a monotonic per-parent sequence. No N+1 patterns were found in the read paths
+sampled.
+
+**Pydantic V2 & Settings.** Fully V2; no `@validator`, `class Config`, `.dict()`, or `parse_obj`
+anywhere. Evidence contracts use `ConfigDict(extra="forbid", frozen=True)` (`providers/evidence.py:47`),
+which is exactly right for persisted provenance — an unexpected field fails loudly rather than
+being silently dropped. `SecretStr` guards the API key. The discriminated
+`SqliteSettings | PostgresSettings` union is clean. `normalize_provider_models` correctly runs
+`mode="before"` so the derived tuple is produced by construction rather than by mutating a frozen
+model — a subtlety that is easy to get wrong. Sole issue: the cwd-coupled `env_file` (MED-03).
+
+**Asyncio Workers.** Notably careful. `asyncio.shield` wraps both the per-page commit and the
+terminal commit (`workflows.py:601-614`, `650-663`), with the `except CancelledError: await task;
+raise` pattern that actually completes the shielded work rather than merely deferring cancellation —
+a detail most implementations get wrong. `handle_worker_exceptions` (`worker.py:157-182`)
+distinguishes retriable from non-retriable faults and stops the loop rather than spinning.
+`_advance_job_with_containment` (`worker.py`/`workflows.py:507-540`) guarantees a claimed job
+cannot strand in `PROCESSING`. `worker_consumer_lifespan` has a bounded shutdown with escalation to
+`cancel()`. Gaps are MED-04 and MED-05, both latent and both with stated unblocking conditions.
+
+**OpenRouter / Adapter Boundary.** Encapsulation holds: `test_provider_boundaries.py` enforces it,
+and `workflows.py` imports only `providers` abstractions. `_CapturingAsyncClient` is a
+well-judged design — it captures the exact transport body before SDK parsing without altering what
+the SDK consumes, including the streamed case. Timeout construction (`openrouter.py:200-206`)
+correctly overrides httpx's 5s per-phase default that would otherwise silently cap the configured
+budget. `SAFE_RESPONSE_HEADERS` (`providers/evidence.py:29-41`) was reviewed field-by-field:
+all nine entries are non-secret correlation, content, or rate-limit headers, and
+`filter_safe_response_headers` is a true allowlist filter with no redaction-after-capture — this
+satisfies invariant 3.8.2 exactly. `_replace_embedded_media` correctly substitutes a source
+reference for base64 payloads, satisfying 3.8.3. The one structural weakness is MED-05.
+
+**Testing & Quality Tooling.** 410 tests, all green, with genuinely strong contract guards
+(boundaries, model contract, media path safety, evidence append-only, atomicity). Marker strictness
+and `asyncio_mode = "strict"` are configured, and no unawaited-coroutine warnings appeared. Two
+guards, however, assert meaningfully less than their names and docstrings claim
+(`test_orphan_sweep.py` — MED-01; `test_errors.py` path-leak coverage — MED-07), and the
+`>= 420` snapshot threshold at `test_orphan_sweep.py:169` repeats a weak-assertion pattern the
+2026-08-23 review already flagged at `>= 200`; it was raised rather than replaced with
+set-membership.
+
+---
+
+## 7. Duplication & Consolidation Report
+
+| Pattern / Duplication | Locations | Proposed Canonical Home | Estimated Lines Removed |
+| :--- | :--- | :--- | :--- |
+| `f"{type(exc).__name__}: {exc}"` detail construction | `errors.py:118`, `maintenance.py:71,95,135,206`, `runtime_settings_store.py:388,476,554` | `errors.py::exception_detail(exc)` | ~8 (consistency > line count) |
+| Filesystem `AppError` construction from `OSError` | `prompts.py:186-191`, `runtime_settings_store.py:384-389,472-477,550-555` | `errors.py::filesystem_error(message, exc, *, suggestion)` | ~20 |
+| Overlapping prompt error types | `services/errors.py:15`, `services/prompts.py:19` | `services/errors.py` (LOW-03) | ~5 |
+| Duplicated `provider_duration_ms` / `processing_duration_ms` max-clamp arithmetic | `workflows.py:341-345, 361-368, 397-404` | `workflows.py::_page_durations(started_at, finished_at, monotonic_started_at)` | ~20 |
+| `_utc_now_naive` defined per module | `workflows.py:51`, `jobs.py:29`, `db/models.py`, `sources.py` | Single helper in `db/models.py`, imported | ~12 |
+
+### Proposed Canonical Abstractions
+
+```python
+# src/transcription/errors.py
+def exception_detail(exc: BaseException) -> str:
+    """Internal-only root-cause text for AppError.detail. Never user-facing."""
+
+
+def filesystem_error[E: AppError](error_type: type[E], message: str, exc: OSError, *, suggestion: str) -> E:
+    """Build a filesystem AppError with a generic message and the path on detail."""
+
+
+# src/transcription/ui/components/error_presenter.py
+def display_failure_detail(error_detail: str | None) -> str | None:
+    """Sanitize persisted failure detail for UI rendering (HIGH-01)."""
+```
+
+---
+
+## 8. Meta-Tooling & Instruction Update Recommendations
+
+1. **`docs/invariant/ai_evidence_and_provenance.md:130`** — resolve the V4 label. Two viable
+   routes, and the maintainer has proposed the second:
+   - **(a) Repair the guard.** Fix the text to V6.1 and make
+     `test_canonical_docs_declare_one_consistent_baseline` token-based rather than phrase-based
+     (MED-02). Keeps version labels as navigational anchors.
+   - **(b) Remove version labels from canonical docs.** While the project has a single principal
+     user and no released versions to support, "canonical" and "current" are the same thing, so the
+     label carries no information a reader can act on — it only creates a second thing to keep in
+     sync. Retain the baseline declaration in `docs/index.md` alone as the release marker, keep
+     version language in `docs/roadmap_plan.md` and the migration/deployment docs (already
+     excluded from the scan for exactly this reason), and replace in-body references with
+     unversioned phrasing ("the canonical architecture, schema, requirements, and error-policy
+     documents"). The guard then inverts: assert that no canonical doc outside the exclusion set
+     contains a version token at all, which is a stricter and much cheaper property to hold than
+     agreement between many labels. Requirement IDs (`REQ-4-021`, `REQ-6-014`) are stable
+     identifiers, not currency claims, and should be left alone.
+2. **`docs/error_handling.md:107-115`** — either add the sanitizing-projection rule for UI display
+   of `error_detail`, or revise the `detail` "Surfaces" row to admit operator-facing evidence
+   displays. Update the "Enforced by" line once MED-07's guard lands, since it currently overstates
+   coverage.
+3. **`.github/instructions/error-handling.instructions.md`** — add an explicit clause under
+   "User-Safe Messaging" stating that *persisted* `error_detail` is subject to the same no-paths
+   rule at any render boundary. The current table (line 86) states the rule for `AppError.detail`
+   and stops there, so the persisted-then-rendered path falls between the lines.
+4. **`.github/instructions/providers.instructions.md`** — record the adapter non-reentrancy
+   constraint (MED-05); it is currently an undocumented precondition of `workflows.py`.
+5. **`.github/instructions/services.instructions.md`** — the two competing atomicity invariants are
+   well described and both guards verified; no change needed. Worth adding the stale-reclaim
+   threshold constraint (MED-04) alongside them, since it is a third worker-lifecycle rule with no
+   documented home.
+6. **`tests/test_orphan_sweep.py`** — repair per MED-01 and replace the `>= 420` threshold with
+   set-membership assertions.
+7. **New `tests/test_error_message_safety.py`** — AST guard per MED-07, covering both the raise
+   sites and the UI render sites.
+8. **`docs/production-runbook.md`** — document the cwd/`WORKDIR` contract for `.env.production`
+   resolution if MED-03 is resolved by documentation rather than by a code seam.
+
+---
+
+## 9. Prioritized Dependency-Ordered Action Plan
+
+**Phase 1 — Blocking fixes (privacy; ordered, HIGH-01 first)**
+1. **HIGH-01** — add `display_failure_detail` and route `sources_page.py:484`,
+   `table/sources.py:90-95`, and `settings_page.py:562` through it. Do this **first**: it closes
+   the render boundary, so the Phase-1.2 fix cannot relocate a leak again.
+2. **HIGH-02** — move paths from `message` to `detail` at the five sites, including the
+   `prompts.py:186` double violation.
+3. **MED-06** — move provider payload text to `detail`; add `detail=` to all three
+   `handle_transcription_errors` handlers.
+
+**Phase 2 — Enforcement hardening (make Phase 1 permanent)**
+4. **MED-07** — AST guard for raise-site `message` safety **and** for UI reads of `error_detail`.
+5. **MED-01** — qualify orphan definitions by module; add module-reachability; replace the
+   snapshot threshold.
+6. **MED-02** — fix the V4/V6.1 text and extend the meta-contract guard to baseline-version currency.
+
+**Phase 3 — Reliability & concurrency (latent; each must precede its unblocking change)**
+7. **MED-04** — derive `worker_stale_job_seconds` from `worker_provider_timeout_seconds` with a
+   rejecting validator, ideally plus a progress heartbeat. **Must land before any second worker
+   replica.**
+8. **MED-05** — scope provider evidence to the call rather than the instance. **Must land before
+   any intra-job page concurrency.** Document non-reentrancy immediately as an interim step.
+
+**Phase 4 — Consolidation & refactoring**
+9. **MED-03** — explicit env-file resolution seam shared by `Settings` and `runtime_settings_store`;
+   remove the `model_config` monkeypatch from `conftest.py`.
+10. **LOW-01** — delete `tags_page.py` (after MED-01, so the suite reproduces the finding).
+11. **LOW-03** and the §7 consolidations — fold in opportunistically while Phase 1 touches these files.
+
+**Phase 5 — Non-blocking governance/documentation depth**
+12. **LOW-02** — relocate or annotate `benchmarking.py`.
+13. Instruction/doc updates §8.3–§8.5, §8.8.
+
+---
+
+## 10. Preserved Strengths
+
+- **Append-only evidence is real, not aspirational.** Every provider call produces a distinct
+  `ExecutionAttempt`; no runtime path mutates a historical row. Projection writes onto
+  `Source.raw_transcription` are clearly separated from history, and `promote_machine_attempt`
+  (`evidence.py:117-146`) repoints the projection without rewriting evidence — with a docstring
+  that explains exactly why that one write lives in a read-oriented service.
+- **Transport-layer terminology is honored in code.** `_CapturingAsyncClient` exists specifically so
+  the stored body is the application-boundary capture rather than an SDK-parsed object, and
+  `TransportEvidence(response_received=False)` explicitly represents "no response" instead of
+  conflating it with an empty one. This is invariant 3.4/3.5 implemented rather than asserted.
+- **Header allowlisting is done the hard, correct way** — filter-before-store with an explicit
+  frozenset, never capture-then-redact (`providers/evidence.py:29-41,130-134`).
+- **Boundaries are enforced by allowlist, not blocklist.** `test_ui_boundaries.py:20-25` states the
+  reasoning explicitly; it means a newly added persistence helper cannot slip through under an
+  unlisted name.
+- **The two competing atomicity invariants are both correctly implemented and both genuinely
+  guarded**, with the tests structured so that the naive over-correction fails.
+- **Cancellation safety in the worker is unusually well handled** — `asyncio.shield` plus
+  `await task` on `CancelledError` actually completes the commit rather than merely deferring
+  cancellation.
+- **Comments explain rationale, not mechanics.** `workflows.py:269-271`, `openrouter.py:200-202`,
+  `config.py:114-115`, and `jobs.py:191-197` each record *why* a non-obvious choice was made,
+  several citing the review log entry that motivated it. This is what made verifying the atomicity
+  and timeout invariants tractable in this review.
+- **Documentation-to-code traceability is strong overall.** Page contracts, schema field tables,
+  and requirement IDs are maintained and guarded; the drift found in this review is narrow and
+  specific rather than systemic.
+
+---
+
+## Appendix A — Repo-Specific Deterministic Checks
+
+| # | Check | Result | Evidence |
+| :-- | :--- | :--- | :--- |
+| 1 | Service boundary rule: no service-to-service imports | **Pass** | `tests/test_service_boundaries.py` green; AST scan, allowlist-based; `workflows.py` composes via `ServiceBundle` |
+| 2 | UI boundary rule: no persistence access from pages/components | **Pass (structurally)** | `tests/test_ui_boundaries.py` green. Caveat: it guards *data access*, not presentation of internal-only fields — see HIGH-01 |
+| 3 | Status vocabulary conformance; no stringly-typed literals | **Pass** | `tests/test_model_contract_guards.py` green; enum members verified against `db/models.py` |
+| 4 | Evidence ownership: append-only history, projections not history mutation | **Pass** | `test_v42_evidence.py::test_attempts_are_append_only_and_exported_with_integrity` verified non-vacuous (asserts both retained attempts and export integrity at lines 281-290) |
+| 5 | Canonical authority: findings resolve against `docs/*` first | **Pass with defect** | `test_canonical_authority_references_are_present` green. The companion baseline-currency guard (`test_canonical_docs_declare_one_consistent_baseline`) scans the offending file but fails open on its phrasing and on minor-less version tokens — MED-02 |
+| 6 | Schema contract fidelity: `docs/schema.md` field-accurate | **Pass** | `test_model_contract_guards.py` + `test_meta_contract_guards.py` green |
+| 7 | Media boundary: record-validated media, controlled URL resolver | **Pass** | `test_media_path_safety.py`, `tests/ui/test_media_urls.py` green; `public_media_path_label` verified path-safe |
+| 8 | Eager-loading conformance vs `lazy="raise"` | **Pass** | Declaration-side guard green; sampled read paths declare explicit `selectinload` chains. Note: dead `tags_page.py:71-74` would violate it if re-registered (LOW-01) |
+| 9 | Cross-cutting error conformance | **FAIL** | Guards green but coverage is narrower than documented: HIGH-01, HIGH-02, MED-06, MED-07 |
+| 10 | Orphan/dead-code conformance | **FAIL** | Guard green but structurally unable to detect a dead module: MED-01, proven by LOW-01 |
+
+## Appendix B — Evidence & Provenance Auditor Families
+
+| Family | Subject | Result | Evidence |
+| :--- | :--- | :--- | :--- |
+| A | Attempt history append-only | **Pass** | No update/delete path to `ExecutionAttempt`; insert-only with `begin_nested` + bounded sequence retry (`sources.py:556-616`) |
+| B | Attempt numbering monotonic per source | **Pass** | `insert_with_sequence_retry`; uniqueness constraint plus retry on conflict |
+| C | Transport evidence captured at the transport boundary | **Pass** | `_CapturingAsyncClient` retains the exact wire body pre-SDK-parse (`openrouter.py:66-94`) |
+| D | Absent response distinguished from empty response | **Pass** | `TransportEvidence.response_received` is explicit, not inferred |
+| E | Response header persistence is allowlist-based | **Pass** | `SAFE_RESPONSE_HEADERS` (`providers/evidence.py:29-41`) — all nine entries verified non-secret; filter-before-store |
+| F | No machine-local detail on user-facing surfaces | **FAIL** | `error_detail` rendered verbatim at three UI sites (HIGH-01); paths in `message` at five sites (HIGH-02) |
+| G | Request manifest excludes embedded media payloads | **Pass** | `_replace_embedded_media` (`openrouter.py:377-395`) substitutes a source reference for base64 data |
+| H | Evidence attribution is correct under concurrency | **Pass today / at risk** | Correct in the current sequential single-worker deployment; the contract itself is non-reentrant (MED-05) and reclaim has no margin (MED-04) |

docs/reviews/README.md

@@ -0,0 +1,21 @@
+# Review Reports
+
+Dated architecture and code review reports generated by
+`.github/skills/python-code-reviewer/skill.md`.
+
+**These files are not canonical authority.** Everything in `docs/reviews/**` is a
+point-in-time observation, not a contract. Canonical intent lives in `docs/index.md`,
+`docs/architecture.md`, `docs/requirements.md`, `docs/schema.md`,
+`docs/error_handling.md`, and `docs/invariant/**`. When a report and a canonical
+document disagree, the canonical document wins until it is deliberately updated.
+
+Naming: `<YYYY-MM-DD>-code-review.md` for review reports, and
+`<YYYY-MM-DD>-remediation-handoff.md` for the implementation plan derived from one.
+
+## Current
+
+- [`2026-08-23-code-review.md`](./2026-08-23-code-review.md) — full review. 0 critical,
+  4 high, 5 medium, 11 low. **All findings remediated.** Retained as a record of the
+  reasoning, not as a list of open work. Note that a few of its recommendations were
+  wrong on contact and were corrected during implementation; the code and the guard
+  tests are authoritative over the report text.

docs/roadmap_plan.md

@@ -0,0 +1,210 @@
+# Roadmap Plan (Starting at V6.0)
+
+This roadmap starts at **V6.0** and tracks forward-looking work only.
+
+## V6.0 - Hosting Migration
+
+Objective: move from local-only operation to secure, stable remote hosting.
+Status: **Completed**
+
+Detailed plan: [`v6_0_hosting_migration_plan.md`](v6_0_hosting_migration_plan.md)
+
+### Scope
+1. Containerize app runtime for production deployment.
+2. Run PostgreSQL in Docker and migrate from SQLite.
+3. Add Cloudflare Tunnel exposure with Access protection.
+4. Add operational safeguards (health checks, restart policies, backups).
+
+### Deliverables
+- Production-ready `docker-compose` deployment for app + database + tunnel.
+- Environment-based configuration for DB, uploads, prompts, and logging.
+- Verified data migration path into PostgreSQL.
+- Runbook updates for deploy, rollback, and backup/restore.
+
+### Exit Criteria
+- `/healthz` reports healthy app and worker in deployed environment.
+- One end-to-end document -> source -> job workflow succeeds remotely.
+- Backup and restore procedure is tested.
+
+### Accomplished
+1. Delivered production Docker deployment with split `app`/`worker`, `postgres`, and `cloudflared`.
+2. Landed SQLite -> PostgreSQL migration tooling and runbook coverage.
+3. Added production health/reliability wiring and operational runbooks for deploy/rollback/recovery.
+4. Established host-visible backup workflow and restore path for PostgreSQL plus media/config assets.
+
+## V6.1 - Testing and Refinement
+
+Objective: improve navigation and operational workflows after user feedback.
+Status: **Completed**
+
+### Scope
+1. Make Document Detail the primary source-page workspace:
+   - Use Source-style pan/zoom + previous/next page controls.
+   - Move editable revision controls into Document Detail.
+   - Move archival/system metadata to dedicated Document Info route.
+2. Simplify top navigation:
+   - Remove top-level Tags and Sources entries.
+   - Retire the Tags page and the global Source Asset Records entry flow.
+3. Improve list/detail clarity:
+   - Add Document transcription status to Archival Documents list.
+   - Add Document Date in People Detail -> Linked Documents table.
+4. Add worker-backed Settings maintenance runs:
+   - Add `maintenance_run` persistence (`id`, `job_type`, `status`, `started_at`, `finished_at`, `triggered_by`, `summary`, `log_path`, `error_detail`).
+   - Add Run Backup and Run Storage Reconciliation actions that enqueue runs and execute in the worker.
+   - Add run history with status, duration, summary, and log view/download.
+   - Defer daily/weekly scheduling controls to V6.2.
+
+### Accomplished
+1. Refactored Document Detail into the primary source-page workspace (pan/zoom viewer, previous/next page navigation, editable revision flow) and moved archival/system metadata to Document Info.
+2. Simplified top navigation by removing Tags/Sources entries and retiring the Tags page/global Source Asset Records flow.
+3. Improved data clarity with document transcription status in Archival Documents and Document Date in People Detail linked documents.
+4. Implemented queue-backed maintenance operations (`maintenance_run` model/service/worker/UI) with run history and log view/download.
+5. Hardened runtime settings operations in production:
+   - runtime settings writes target mounted `.env.production`,
+   - fallback write path for single-file bind mounts,
+   - explicit hidden/deployment-key disclosure in Settings UI.
+6. Simplified backup configuration and behavior:
+   - standardized on `BACKUP_DIR` + `BACKUP_RETENTION_DAYS`,
+   - backup script uses `DATABASE__*` persistence keys,
+   - compose maps Postgres container init values from `DATABASE__*`,
+   - env contract drift tests now guard `.env.production.example`.
+
+## V6.2 - GEDCOM Data Layer
+
+Objective: introduce a genealogical data layer sourced from GEDCOM exports, bridged to
+existing `Person` records via FamilySearch ID, without disrupting document-focused Person
+workflows.
+
+### Scope
+1. Manual `.ged` file upload only. No FamilySearch credentials are stored or used by the
+   app; the user runs the third-party `getmyancestors` tool themselves and uploads the
+   resulting export.
+2. Four new tables: `genealogy_person`, `genealogy_family`, `genealogy_family_child`, and
+   `genealogy_citation` (raw GEDCOM `SOUR` citations, reusable in a later version to record
+   when a transcribed document itself becomes citation evidence for FamilySearch).
+3. Upsert-based import keyed on FamilySearch ID (`fs_id`) so repeat imports update existing
+   records in place without breaking existing `Person.family_search_id` links or duplicating
+   surrogate keys.
+4. Reuse the existing V6.1 worker-backed `maintenance_run` pattern for import runs (run
+   history, status, summary, log view/download) rather than new infrastructure.
+
+### Deliverables
+- GEDCOM parser/importer producing the four genealogy tables.
+- `MaintenanceJobType` entry for GEDCOM import with upsert semantics and a run summary
+  (records added/updated).
+- Settings UI entry to upload a `.ged` file, trigger an import run, and view history.
+
+### Exit Criteria
+- Importing the same `.ged` file twice does not duplicate or orphan data.
+- Existing `Person.family_search_id` values continue to resolve to the correct
+  `genealogy_person` row after import.
+- Import run history is visible with status, duration, and summary, consistent with other
+  maintenance runs.
+
+## V6.3 - Reporting and Genealogy-Enriched Features
+
+Objective: improve research value with person-centric outputs, grounded in both archival
+documents and the V6.2 genealogical data layer.
+
+This version is broken into five sequential sub-versions because of real dependency
+ordering: entity linking must exist before GEDCOM data can be targeted per-person; the
+Facts/Events mechanism must exist before timelines or reconciliation have anything
+meaningful to consume.
+
+### V6.3.1 - Manual Entity Linking
+
+- Search/browse UI over `genealogy_person` to find and link a candidate match to an
+  application `Person`, setting `family_search_id`. Linking is reversible (unlink).
+- Once linked, GEDCOM vitals display alongside the `Person` record without requiring any
+  schema change to `Person`.
+
+### V6.3.2 - Person Facts and Events
+
+- New fact/event table capturing: person, fact type (birth/death/event/free-form), date
+  (+raw), place, free-text description, and a link to the source document as evidence.
+- Manual tagging UI while reviewing a transcribed document: select a passage, choose the
+  person and fact type, record the date/description.
+- One-time migration of existing `Person.birth_date`/`birth_date_raw`/`birth_place`/
+  `death_date`/`death_date_raw`/`death_place` values into fact/event rows (tagged as
+  legacy/no-document-evidence where no source document is known), followed by retiring those
+  six columns from `Person`. Birth/death become Facts/Events like any other locally-known
+  fact, for both linked and unlinked people. `Person` permanently keeps `last_name`,
+  `given_names`, `biography`, `family_search_id`, `metadata_`, tags, photos, and document
+  associations.
+
+### V6.3.3 - Person Timelines
+
+- Timeline query merging GEDCOM milestones (birth, marriage, children's births, death) for
+  linked persons with locally recorded Facts/Events.
+- Timeline UI on Person Detail with clear ordering/filters; entries link back to their
+  originating document or GEDCOM record.
+
+### V6.3.4 - Reconciliation
+
+- Compares Facts/Events (the real, document-evidenced local signal) against corresponding
+  `genealogy_person` fields for linked persons.
+- Persisted reconciliation record: person, field, local value with evidence-document link,
+  GEDCOM value, and status (open / submitted / dismissed).
+- Re-evaluated automatically as part of each GEDCOM import maintenance run: opens new
+  discrepancies, auto-resolves ones where GEDCOM now matches, leaves others unchanged.
+- Reconciliation review UI functions as a manual to-do list for updating FamilySearch; the
+  app does not write back to FamilySearch itself.
+
+### V6.3.5 - AI-Assisted Biography Generation
+
+- Prompted narrative generation grounded in GEDCOM facts, Facts/Events, and relevant
+  document snippets as structured input, using existing evidence-safe prompting patterns.
+- Output cites back to source documents and FamilySearch records.
+- Saved/printable report presentation for review; reports do not modify archival source
+  data.
+
+### Exit Criteria (applies across V6.3.1-V6.3.5)
+- Entity links are reversible and do not alter document associations.
+- Timelines are reproducible from persisted records.
+- Reconciliation items always carry a link to the document evidence justifying the local
+  value, and re-running GEDCOM import correctly opens, resolves, or leaves items unchanged.
+- Narrative generation is traceable to source records and prompts.
+- Reports can be reviewed without modifying archival source data.
+
+## V6.4 - Access Control and Multi-User Readiness
+
+[ *More thoughts on user accounts:*
+* *Create a generic "view only" user that does not have the rights to alter any of the data*
+* *Limit user accounts access to data by Tag. I have distant family members that I would want to share the transcribed data with, but they would only be interested in a subset of it. For example my Cochran cousins would have no interest in Lancaster documents, so limit the Cochra Clan cousins to view-only access to documents tagged "cochran clan"* ]
+
+Objective: prepare for managed collaboration beyond single-user operation.
+
+### Scope
+1. Introduce application-level authentication.
+2. Add role-based authorization (admin/editor/contributor/viewer).
+3. Add audit visibility for user-attributed write actions.
+
+### Deliverables
+- User identity model and login/session flow.
+- Route/page/service authorization enforcement.
+- Audit metadata for sensitive create/update/delete workflows.
+
+### Exit Criteria
+- Unauthorized operations are blocked consistently across UI/API.
+- Role policies are enforced by deterministic tests.
+- User-attributed changes are visible for audit/review.
+
+## Deferred / Future Ideas (not committed scope)
+
+Captured for later consideration, not yet scheduled to a version:
+* AI-assisted entity disambiguation (kinship co-occurrence, chronological plausibility
+  filtering) when linking document mentions to people.
+* Kinship-aware `@mention` tagging while transcribing.
+* Relationship-calculator badges (e.g., "3rd Great-Grandmother") in the document viewer.
+* Interactive migration/geography mapping from GEDCOM and document place mentions.
+* AI-suggested document discovery by date/location overlap with known persons.
+* Ability to search within a document to find potential people to add to the People table.
+
+## Planning Notes
+
+- Keep architecture, schema, and UI contracts synchronized in `docs/` as each version lands.
+- Prefer explicit schema migration over runtime compatibility write paths.
+- Preserve evidence/provenance guarantees when adding new AI-powered features.
+- GEDCOM/FamilySearch data is external, collaborative, and mutable; treat it as a managed
+  cache bridged via `fs_id`, never as a replacement for archival evidence recorded from
+  transcribed documents.

docs/schema.md

@@ -0,0 +1,391 @@
+# Data Model and Persistence Schema (Current Baseline: V6.1)
+
+This document is the field-accurate V6.1 schema contract aligned to `src/transcription/db/models.py`.
+
+## Source of Truth Anchors
+
+- `src/transcription/db/models.py` (status and purpose enums, including maintenance lifecycle enums)
+- `src/transcription/db/models.py:80-120` (`DocumentType`, `PersonRole`)
+- `src/transcription/db/models.py:122-172` (`Tag`, `Document`)
+- `src/transcription/db/models.py` (`Person`, `GenealogyPerson`, `GenealogyFamily`, `GenealogyFamilyChild`, `GenealogyCitation`)
+- `src/transcription/db/models.py` (`Photo`, `DocumentPerson`, `DocumentTag`)
+- `src/transcription/db/models.py:285-347` (`Job`)
+- `src/transcription/db/models.py` (`MaintenanceRun`)
+- `src/transcription/db/models.py:350-462` (`Source`, `JobSource`)
+- `src/transcription/db/models.py:465-522` (`ExecutionAttempt`)
+
+## Entity Relationship Overview
+
+```mermaid
+erDiagram
+    DocumentType ||--o{ Document : classifies
+    Document ||--o{ Job : has
+    Document ||--o{ Source : has
+    Document ||--o{ DocumentPerson : links
+    Document ||--o{ DocumentTag : tagged
+    Person ||--o{ DocumentPerson : links
+    Person ||--o{ PersonTag : tagged
+    Person ||--o{ Photo : owns
+    GenealogyPerson ||--o{ GenealogyFamily : husband
+    GenealogyPerson ||--o{ GenealogyFamily : wife
+    GenealogyPerson ||--o{ GenealogyFamilyChild : child
+    GenealogyFamily ||--o{ GenealogyFamilyChild : includes
+    GenealogyPerson ||--o{ GenealogyCitation : cited
+    GenealogyFamily ||--o{ GenealogyCitation : cited
+    Document ||--o{ GenealogyCitation : evidence
+    PersonRole ||--o{ DocumentPerson : labels
+    Tag ||--o{ DocumentTag : labels
+    Tag ||--o{ PersonTag : labels
+    Job ||--o{ JobSource : includes
+    Source ||--o{ JobSource : participates
+    JobSource ||--o{ ExecutionAttempt : attempts
+    MaintenanceRun {
+        uuid id PK
+    }
+```
+
+## Authoritative Enumerations
+
+### JobStatus
+
+- `queued`
+- `processing`
+- `transcribed`
+- `partial_success`
+- `failed`
+
+### JobSourceStatus
+
+- `pending`
+- `transcribed`
+- `failed`
+- `cancelled`
+
+### JobPurpose
+
+- `transcription`
+- `retranscription`
+
+### MaintenanceJobType
+
+- `backup`
+- `storage_reconciliation`
+- `gedcom_import`
+
+### MaintenanceRunStatus
+
+- `queued`
+- `processing`
+- `succeeded`
+- `failed`
+
+## Field-Accurate Table Contracts
+
+### `DocumentType`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `semantic_key` | `str \| None` | nullable unique, indexed |
+| `label` | `str` | required |
+| `normalized_label` | `str` | unique, indexed |
+| `is_active` | `bool` | default `True` |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `PersonRole`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `semantic_key` | `str \| None` | nullable unique, indexed |
+| `label` | `str` | required |
+| `normalized_label` | `str` | unique, indexed |
+| `is_active` | `bool` | default `True` |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `Tag`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `semantic_key` | `str \| None` | nullable unique, indexed |
+| `label` | `str` | required |
+| `normalized_label` | `str` | unique, indexed |
+| `is_active` | `bool` | default `True` |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `Document`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `name` | `str` | required |
+| `document_type_id` | `UUID \| None` | FK -> `document_type.id`, indexed |
+| `document_date` | `date \| None` | optional |
+| `document_date_raw` | `str \| None` | optional |
+| `location_created` | `str \| None` | optional |
+| `notes` | `str \| None` | optional |
+| `archive_identifier` | `str \| None` | optional |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `Person`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `last_name` | `str` | required |
+| `given_names` | `str` | required |
+| `birth_date` | `date \| None` | optional |
+| `birth_date_raw` | `str \| None` | optional |
+| `birth_place` | `str \| None` | optional |
+| `death_date` | `date \| None` | optional |
+| `death_date_raw` | `str \| None` | optional |
+| `death_place` | `str \| None` | optional |
+| `biography` | `str \| None` | optional |
+| `family_search_id` | `str \| None` | nullable unique |
+| `metadata_` | `dict[str, JsonValue] \| None` | stored as DB column `metadata` (`JSONBCompat`) |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `GenealogyPerson`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `fs_id` | `str` | unique, indexed FamilySearch identifier |
+| `full_name` | `str` | required |
+| `birth_date` | `date \| None` | optional |
+| `birth_date_raw` | `str \| None` | optional |
+| `birth_place` | `str \| None` | optional |
+| `death_date` | `date \| None` | optional |
+| `death_date_raw` | `str \| None` | optional |
+| `death_place` | `str \| None` | optional |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `GenealogyFamily`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `fs_family_id` | `str` | unique, indexed FamilySearch family identifier |
+| `husband_id` | `UUID \| None` | nullable FK -> `genealogy_person.id`, indexed |
+| `wife_id` | `UUID \| None` | nullable FK -> `genealogy_person.id`, indexed |
+| `marriage_date` | `date \| None` | optional |
+| `marriage_date_raw` | `str \| None` | optional |
+| `marriage_place` | `str \| None` | optional |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `GenealogyFamilyChild`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `family_id` | `UUID` | FK -> `genealogy_family.id`, indexed |
+| `child_id` | `UUID` | FK -> `genealogy_person.id`, indexed |
+| `relationship_type` | `str \| None` | optional |
+| `created_at` | `datetime` | default now |
+
+Constraint:
+- `UniqueConstraint(family_id, child_id)` named `uq_genealogy_family_child`
+
+### `GenealogyCitation`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `genealogy_person_id` | `UUID \| None` | nullable FK -> `genealogy_person.id`, indexed |
+| `genealogy_family_id` | `UUID \| None` | nullable FK -> `genealogy_family.id`, indexed |
+| `fact_type` | `GenealogyCitationFactType` | enum: `birth`, `death`, `marriage`, `other` |
+| `raw_citation_text` | `str` | required raw GEDCOM citation text |
+| `source_kind` | `GenealogyCitationSourceKind` | enum: `familysearch_imported`, `transcription_evidence` |
+| `document_id` | `UUID \| None` | nullable FK -> `document.id`, indexed |
+| `created_at` | `datetime` | default now |
+
+### `Photo`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `person_id` | `UUID \| None` | nullable FK -> `person.id`, indexed (`NULL` = homepage photo) |
+| `path` | `str` | required upload-root-relative POSIX path (`photos/...`) |
+| `description` | `str \| None` | optional |
+| `is_primary` | `bool` | default `False`; owner-level "featured/primary" marker |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `DocumentPerson`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `document_id` | `UUID` | FK -> `document.id`, indexed |
+| `person_id` | `UUID` | FK -> `person.id`, indexed |
+| `role_id` | `UUID` | FK -> `person_role.id`, indexed |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+Constraint:
+- `UniqueConstraint(document_id, person_id)` named `uq_document_person`
+
+### `DocumentTag`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `document_id` | `UUID` | FK -> `document.id`, indexed |
+| `tag_id` | `UUID` | FK -> `tag.id`, indexed |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+Constraint:
+- `UniqueConstraint(document_id, tag_id)` named `uq_document_tag`
+
+### `PersonTag`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `person_id` | `UUID` | FK -> `person.id`, indexed |
+| `tag_id` | `UUID` | FK -> `tag.id`, indexed |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+Constraint:
+- `UniqueConstraint(person_id, tag_id)` named `uq_person_tag`
+
+### `Job`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `document_id` | `UUID` | FK -> `document.id`, indexed |
+| `status` | `JobStatus` | non-null enum (stored as enum values) |
+| `retry_count` | `int` | default `0`, `ge=0` |
+| `purpose` | `JobPurpose` | non-null enum, default `transcription` |
+| `date_created` | `datetime` | default now |
+| `date_updated` | `datetime` | default now, onupdate |
+| `provider` | `str \| None` | optional |
+| `model` | `str \| None` | optional |
+| `prompt_name` | `str \| None` | optional |
+| `prompt_hash` | `str \| None` | optional |
+| `system_prompt` | `str \| None` | optional |
+| `user_prompt` | `str \| None` | optional |
+| `temperature` | `float \| None` | optional |
+| `top_p` | `float \| None` | optional |
+
+Index:
+- `Index("ix_job_status_date_created", "status", "date_created")`
+
+### `MaintenanceRun`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `job_type` | `MaintenanceJobType` | non-null enum |
+| `status` | `MaintenanceRunStatus` | non-null enum, default `queued` |
+| `started_at` | `datetime \| None` | optional |
+| `finished_at` | `datetime \| None` | optional |
+| `triggered_by` | `str \| None` | optional |
+| `summary` | `str \| None` | optional |
+| `log_path` | `str \| None` | optional, log-root-relative POSIX path |
+| `error_detail` | `str \| None` | optional internal failure detail |
+| `created_at` | `datetime` | default now |
+| `updated_at` | `datetime` | default now, onupdate |
+
+### `Source`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `document_id` | `UUID` | FK -> `document.id`, indexed |
+| `page_number` | `int` | default `1`, `ge=1` |
+| `upload_name` | `str` | required |
+| `filename` | `str` | required |
+| `file_path` | `str` | required upload-root-relative POSIX path (`documents/...`) |
+| `file_hash` | `str` | required |
+| `file_size_bytes` | `int` | `BigInteger`, non-null |
+| `raw_transcription` | `str \| None` | projection field |
+| `preferred_execution_attempt_id` | `UUID \| None` | nullable FK -> `execution_attempt.id`, indexed (`use_alter`) |
+| `revised_text` | `str \| None` | optional human revision |
+| `date_uploaded` | `datetime` | default now |
+| `date_revised` | `datetime \| None` | optional |
+
+### `JobSource`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `job_id` | `UUID` | FK -> `job.id`, indexed |
+| `source_id` | `UUID` | FK -> `source.id`, indexed |
+| `status` | `JobSourceStatus` | non-null enum, default `pending` |
+
+Constraint:
+- `UniqueConstraint(job_id, source_id)` named `uq_job_source_job_source`
+
+Runtime reconciliation:
+- Startup database operations remove retired V4.6 `job_source` evidence columns (`raw_transcription`, `ai_metadata`, `raw_api_response`, `error_detail`, `executed_at`) when present so persisted schema matches this contract.
+
+### `ExecutionAttempt`
+
+| Field | Type | Notes |
+| :--- | :--- | :--- |
+| `id` | `UUID` | PK |
+| `job_source_id` | `UUID` | FK -> `job_source.id`, indexed |
+| `job_id` | `UUID` | FK -> `job.id`, indexed |
+| `source_id` | `UUID` | FK -> `source.id`, indexed |
+| `attempt_number` | `int` | `ge=1` |
+| `status` | `JobSourceStatus` | non-null enum, value-stable with `JobSource.status` |
+| `provider` | `str` | required |
+| `model` | `str \| None` | optional |
+| `request_manifest` | `dict[str, JsonValue] \| None` | JSONBCompat |
+| `request_manifest_sha256` | `str \| None` | optional |
+| `request_manifest_schema_version` | `str \| None` | optional |
+| `response_received` | `bool` | default `False` |
+| `transport_status_code` | `int \| None` | optional |
+| `transport_body` | `bytes \| None` | LargeBinary |
+| `transport_content_type` | `str \| None` | optional |
+| `transport_content_encoding` | `str \| None` | optional |
+| `transport_safe_headers` | `dict[str, JsonValue] \| None` | JSONBCompat |
+| `router_request_id` | `str \| None` | optional |
+| `router_generation_id` | `str \| None` | optional |
+| `sdk_response_snapshot` | `dict[str, JsonValue] \| None` | JSONBCompat |
+| `normalized_metadata` | `dict[str, JsonValue] \| None` | JSONBCompat; may include app-namespaced `processing_timing` (`provider_call_duration_ms`, `processing_duration_ms`) |
+| `software_context` | `dict[str, JsonValue] \| None` | JSONBCompat |
+| `raw_transcription` | `str \| None` | optional |
+| `error_category` | `str \| None` | optional |
+| `error_detail` | `str \| None` | optional |
+| `failure_phase` | `str \| None` | optional |
+| `started_at` | `datetime` | required |
+| `finished_at` | `datetime` | required |
+| `duration_ms` | `int` | `ge=0` |
+| `created_at` | `datetime` | default now |
+
+Constraint:
+- `UniqueConstraint(job_id, source_id, attempt_number)` named `uq_execution_attempt_number`
+
+## Relationship Loading Contract
+
+- Most ORM relationships are configured with `lazy="raise"`.
+- `JobSource.execution_attempts` is intentionally `lazy="noload"` with ordered attempts.
+- Service/UI read paths must explicitly eager-load required relationships before access.
+
+## Persistence Invariants (Ground Truth)
+
+1. `ExecutionAttempt` is append-only runtime evidence.
+2. `JobSource.status` represents queue/projection execution state and is not a full evidence container.
+3. `Source.raw_transcription` is a mutable projection and not authoritative attempt history.
+4. `Job` terminal status derives from page outcomes (`JobSource` state), not from a separate summary table.
+5. `DocumentType.semantic_key` and `PersonRole.semantic_key` are nullable-unique semantic identifiers.
+
+## Cross-Reference
+
+- [System Architecture](architecture.md)
+- [System Requirements](requirements.md)
+- [Error Handling Policy](error_handling.md)
+- [AI Evidence and Provenance Invariant](./invariant/ai_evidence_and_provenance.md)

docs/schema_v2.md

@@ -1,136 +0,0 @@
-# Database Schema (Version 2)
-
-This document describes the PostgreSQL relational schema for the transcription platform. It incorporates multi-image batch orchestration via `asyncio`, page-level execution tracking, many-to-many author/recipient attribution, and JSONB document storage for AI vision outputs.
-
-All primary and foreign keys are PostgreSQL native UUIDs (`gen_random_uuid()`).
-
-## Entity Relationship Diagram
-
-```mermaid
-erDiagram
-    PERSON {
-        UUID id PK
-        TEXT full_name
-        TEXT display_name
-        TEXT maiden_name
-        DATE birth_date
-        TEXT birth_date_raw
-        TEXT birth_place
-        DATE death_date
-        TEXT death_date_raw
-        TEXT death_place
-        TEXT biography
-        TEXT portrait_path
-        JSONB metadata
-        TIMESTAMPTZ created_at
-        TIMESTAMPTZ updated_at
-    }
-
-    DOCUMENT {
-        UUID id PK
-        TEXT name
-        TEXT document_type
-        DATE document_date
-        TEXT document_date_raw
-        TEXT location_created
-        TEXT notes
-        TEXT archive_identifier
-        TIMESTAMPTZ created_at
-        TIMESTAMPTZ updated_at
-    }
-
-    DOCUMENT_PERSON {
-        UUID id PK
-        UUID document_id FK
-        UUID person_id FK
-        VARCHAR role "author | recipient"
-        TIMESTAMPTZ created_at
-    }
-
-    JOB {
-        UUID id PK
-        UUID document_id FK
-        VARCHAR status "queued | processing | completed | partial_success | failed"
-        INTEGER retry_count
-        TEXT provider
-        TEXT model
-        TEXT prompt_name
-        TIMESTAMPTZ date_created
-        TIMESTAMPTZ date_updated
-    }
-
-    SOURCE {
-        UUID id PK
-        UUID document_id FK
-        INTEGER page_number
-        TEXT upload_name
-        TEXT filename
-        TEXT file_path
-        TEXT raw_transcription
-        TEXT revised_text
-        TIMESTAMPTZ date_uploaded
-        TIMESTAMPTZ date_revised
-    }
-
-    JOB_SOURCE {
-        UUID id PK
-        UUID job_id FK
-        UUID source_id FK
-        VARCHAR status "pending | transcribed | failed"
-        TEXT raw_transcription
-        JSONB ai_metadata
-        JSONB raw_api_response
-        TEXT error_detail
-        TIMESTAMPTZ executed_at
-    }
-
-    DOCUMENT ||--o{ DOCUMENT_PERSON : "has_people"
-    PERSON ||--o{ DOCUMENT_PERSON : "participates_in"
-    DOCUMENT ||--o{ JOB : "has_jobs"
-    DOCUMENT ||--o{ SOURCE : "contains_pages"
-    JOB ||--o{ JOB_SOURCE : "executes"
-    SOURCE ||--o{ JOB_SOURCE : "processed_in"
-```
-
-## Domain Invariants & Rules
-
-### Page-Level Execution & AI Outputs
-
-* Execution Granularity: Every single image execution by an AI model produces a dedicated record in job_source.
-* Point-in-Time Auditability: job_source.raw_api_response stores the unparsed REST response envelope for that specific image page call. job_source.ai_metadata stores spatial bounding boxes, token usage, and layout details for that specific image page call.
-* Active Output Caching: Upon successful completion of an image call, source.raw_transcription is updated with the latest output string from job_source.raw_transcription for fast UI rendering.
-
-### Page Ordering & Revisions
-
-* Sequential Integrity: source.page_number dictates page ordering within a document. Reads assembling full documents must query ORDER BY source.document_id, source.page_number ASC.
-* Inlined Human Corrections: User edits occur at the page level inside source.revised_text. source.raw_transcription remains immutable. If source.revised_text is non-null, application frontends must render source.revised_text.
-
-### Async Job Lifecycle & Failure Isolation
-
-* Batch Orchestrator: A job represents an overarching execution run across one or more source images belonging to a document.
-* Isolated Failures: API requests run concurrently (e.g., using asyncio). A failure on page 3 does not invalidate successful transcriptions on page 1 or 2.
-* Job States:
-    - queued: Created, awaiting worker execution.
-    - processing: Concurrent HTTP tasks actively running.
-    - completed: 100% of linked job_source tasks succeeded (transcribed).
-    - partial_success: At least one job_source succeeded and at least one failed.
-    - failed: All linked job_source tasks failed or a job-level runtime error occurred.
-
-### Attribution & Person Roles
-
-* Multi-Person Roles: Documents support zero, one, or many authors and recipients linked via document_person.
-* Role Uniqueness: (document_id, person_id, role) must be unique to prevent duplicate role tagging.
-
----
-
-## Related Local References
-
-- [System Overview](index_v2.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v2.md)
-- [System Requirements](requirements_v2.md)
-- Data model (this document)
-- [Error Handling Policy](error_handling_v2.md)
-- [Implementation Plan](implementation_plan_v2.md)
-

docs/test_block-beta.md

@@ -1,92 +0,0 @@
-```mermaid
-block-beta
-  columns 3
-
-  %% UI Component Column
-  block:UI["UI COMPONENTS / WIREFRAME"]:1
-    columns 1
-    
-    block:HeaderUI["Header & Nav"]:1
-      columns 1
-      h_title["[Text] Document Name & Type"]
-      h_date["[Text] Date & Origin Location"]
-    end
-
-    block:EditorUI["Page Transcription Editor"]:1
-      columns 1
-      ed_img["[Image Viewer] Source Image"]
-      ed_page["[Badge] Page Number"]
-      ed_raw["[Read-Only] AI Raw Output"]
-      ed_rev["[Textarea] Human Revised Text"]
-    end
-
-    block:PeopleUI["Attribution Sidebar"]:1
-      columns 1
-      p_author["[List] Authors (Full Name)"]
-      p_recip["[List] Recipients (Full Name)"]
-      p_bio["[Card] Person Biography & Dates"]
-    end
-
-    block:JobUI["AI Processing Drawer"]:1
-      columns 1
-      j_status["[Badge] Job Status"]
-      j_model["[Text] Provider & Model"]
-      j_tokens["[JSON View] AI Token Usage"]
-    end
-  end
-
-  %% Directional Mapping / Connectors
-  block:FLOW["MAPPING / FLOW"]:1
-    columns 1
-    f1["Reads / Updates -->"]
-    f2["Renders Active Page -->"]
-    f3["Joins via Role -->"]
-    f4["Executes & Logs -->"]
-  end
-
-  %% Postgres Schema Column
-  block:DB["POSTGRES SQL SCHEMA"]:1
-    columns 1
-
-    block:DocTbl["Table: document"]:1
-      columns 1
-      d_id["id : UUID (PK)"]
-      d_name["name : TEXT"]
-      d_type["document_type : TEXT"]
-      d_date["document_date : DATE"]
-    end
-
-    block:SrcTbl["Table: source"]:1
-      columns 1
-      s_id["id : UUID (PK)"]
-      s_page["page_number : INT"]
-      s_path["file_path : TEXT"]
-      s_raw["raw_transcription : TEXT"]
-      s_rev["revised_text : TEXT"]
-    end
-
-    block:PersonTbl["Table: person & document_person"]:1
-      columns 1
-      p_id["id : UUID (PK)"]
-      p_name["full_name : TEXT"]
-      p_role["role : 'author' | 'recipient'"]
-    end
-
-    block:JobTbl["Table: job & job_source"]:1
-      columns 1
-      j_id["id : UUID (PK)"]
-      j_stat["status : VARCHAR"]
-      j_prov["provider / model : TEXT"]
-      j_meta["ai_metadata : JSONB"]
-    end
-  end
-
-  %% Connections
-  HeaderUI --> DocTbl
-  ed_img --> s_path
-  ed_page --> s_page
-  ed_raw --> s_raw
-  ed_rev --> s_rev
-  PeopleUI --> PersonTbl
-  JobUI --> JobTbl
-```

docs/test_flowchart.md

@@ -1,53 +0,0 @@
-```mermaid
-flowchart LR
-    subgraph UI["UI Components / Wireframe"]
-        direction TB
-        subgraph HeaderUI["Header & Nav"]
-            h_title["[Text] Document Name & Type"]
-            h_date["[Text] Date & Origin Location"]
-        end
-        subgraph EditorUI["Page Transcription Editor"]
-            ed_img["[Image Viewer] Source Image"]
-            ed_page["[Badge] Page Number"]
-            ed_raw["[Read-Only] AI Raw Output"]
-            ed_rev["[Textarea] Human Revised Text"]
-        end
-        subgraph PeopleUI["Attribution Sidebar"]
-            p_author["[List] Authors / Recipients"]
-        end
-        subgraph JobUI["AI Processing Drawer"]
-            j_status["[Badge] Job Status"]
-        end
-    end
-
-    subgraph DB["Postgres SQL Schema"]
-        direction TB
-        subgraph DocTbl["Table: document"]
-            d_name["name : TEXT"]
-            d_type["document_type : TEXT"]
-        end
-        subgraph SrcTbl["Table: source"]
-            s_path["file_path : TEXT"]
-            s_page["page_number : INT"]
-            s_raw["raw_transcription : TEXT"]
-            s_rev["revised_text : TEXT"]
-        end
-        subgraph PersonTbl["Table: person & document_person"]
-            p_name["full_name : TEXT"]
-            p_role["role : author | recipient"]
-        end
-        subgraph JobTbl["Table: job & job_source"]
-            j_stat["status : VARCHAR"]
-            j_meta["ai_metadata : JSONB"]
-        end
-    end
-
-    %% Mappings
-    HeaderUI --> DocTbl
-    ed_img --> s_path
-    ed_page --> s_page
-    ed_raw --> s_raw
-    ed_rev --> s_rev
-    PeopleUI --> PersonTbl
-    JobUI --> JobTbl
-```

docs/ui/README.md

@@ -0,0 +1,60 @@
+# UI Behavioral Contracts
+
+## Purpose
+
+This directory defines the current user-facing behavior of the NiceGUI application. It records what each page is for, which routes and actions it exposes, what information it presents, and how success, empty, validation, and failure states behave.
+
+These documents are written for maintainers and AI contributors. They are behavioral contracts, not historical implementation notes and not substitutes for the database schema.
+
+## Current Page Contracts
+
+- [Home](pages/home.md)
+- [Documents](pages/documents.md)
+- [People](pages/people.md)
+- [Jobs](pages/jobs.md)
+- [Sources](pages/sources.md)
+- [Settings](pages/settings.md)
+
+NiceGUI registers the routes shown in each contract without the `/ui` prefix. The application mounts NiceGUI under `/ui`, so `/documents` in page code is served to a browser as `/ui/documents`.
+
+## Authority Hierarchy
+
+When documents disagree, use this order:
+
+1. User-facing page intent and accepted behavior: the page contracts in this directory.
+2. Visual and interaction styling: [UI Style Guide](../invariant/ui_style_guide.md).
+3. UI dependency and ownership boundaries: [UI contributor instructions](../../.github/instructions/ui.instructions.md).
+4. Durable failure behavior: [Error Handling invariant](../invariant/error_handling.md).
+5. Durable AI evidence behavior: [Digital Evidence and AI Processing Provenance](../invariant/ai_evidence_and_provenance.md).
+6. Data definitions and relationships: current models plus the [schema contract](../schema.md).
+7. Implementation truth: current code and tests.
+
+If code intentionally changes accepted page behavior, update the corresponding page contract in the same change. If code accidentally differs, correct the implementation rather than rewriting intent to match a defect.
+
+## Contract Contents
+
+Each page contract contains:
+
+1. Purpose and user goals.
+2. Registered routes and navigation context.
+3. List, detail, and form behavior.
+4. Editable and system-managed information.
+5. Validation, empty, loading, and failure states.
+6. A concise acceptance checklist.
+7. Current implementation and test anchors.
+8. Known limitations and deferred work.
+
+## Maintenance Rules
+
+- Describe current accepted behavior in present tense.
+- Do not mix an obsolete “first release” design with current behavior.
+- Keep future changes in versioned scope documents and link to them from a Deferred Work section.
+- Do not reproduce the complete database field inventory here; include only fields that affect page behavior.
+- Keep service, file, and test anchors current.
+- Do not create separate current-state, target-state, and traceability copies of the same contract.
+- Keep cross-page visual rules in the UI Style Guide instead of repeating them on each page.
+- Keep database joins such as `DocumentPerson` and `JobSource` in schema/architecture documentation unless they directly affect a page interaction.
+
+## Current Baseline
+
+These contracts describe the current V6.1 baseline.

docs/ui/pages/documents.md

@@ -0,0 +1,150 @@
+# Documents Page Contract
+
+## Purpose
+
+Documents manages the archival record for each historical artifact independently of its source files and transcription jobs. A Document can be created first, linked to people in one or more roles, and used later as the parent for Sources and Jobs.
+
+## Routes
+
+| Route | Purpose |
+| --- | --- |
+| `/documents` | Searchable archival Document list. |
+| `/documents/new` | Create a Document. |
+| `/documents/{document_id}` | View one Document and its related records. |
+| `/documents/{document_id}/info` | View archival metadata and system logistics for one Document. |
+| `/documents/{document_id}/edit` | Edit metadata and the complete Linked People set. |
+| `/documents/{document_id}/delete` | Confirm or block deletion. |
+| `/documents/{document_id}/jobs` | Show Jobs belonging to the Document. |
+| `/documents/{document_id}/sources` | Source-image gallery for the Document. |
+| `/documents/{document_id}/print` | Preview and browser-print the persisted Document. |
+
+## List Behavior
+
+- The title is **Archival Documents**.
+- **Create new document** opens the create route.
+- The table defaults to Document Title order and supports search and column sorting.
+- Columns are Document Title, Author, Tags, Document Date, Type, # Sources, and Transcription Status.
+- Document Title is left-aligned; the remaining columns are centered.
+- Author lists all linked people in the `author` role.
+- # Sources reflects the count of linked Source rows for each Document.
+- Transcription Status reflects the most recent Job status for that Document; documents with no Jobs show a blank marker.
+- Date display prefers exact date, then approximate date, then `Unknown`.
+- Selecting a row opens Document Detail.
+- Row navigation includes list context so Document Detail provides **Back to Documents**.
+- No records displays `No documents found in repository.`
+
+## Create and Edit Behavior
+
+Required:
+
+- Document name.
+- Document type selected from the Document Type registry.
+
+Optional:
+
+- Exact date.
+- Approximate date.
+- Document location.
+- Archive identifier.
+- Notes.
+- Tags.
+- Linked People, with exactly one Person Role per linked Person.
+
+Rules:
+
+- Exact date must parse as `YYYY-MM-DD`; browser presentation may follow locale.
+- The exact-date input is labeled **Document date**.
+- Existing people appear with disambiguating labels.
+- Tag assignment supports selecting existing tags and adding new labels inline.
+- **Create new person** opens Person creation.
+- `person_id` may preselect that Person in the author role on Document creation.
+- An invalid requested Person produces a warning rather than a broken form.
+- `return_to=jobs_new` returns a successful create to Job creation with the new Document selected.
+- Edit includes active and inactive Document Types so historical values remain maintainable.
+- One Linked People table contains Select, Person, and Role columns.
+- Add and Edit use an inline Person/Role editor; Save, Cancel, and Delete change staged UI state only.
+- A Person may appear once per Document regardless of role.
+- Existing inactive-role links remain visible; only active roles may be newly assigned.
+- Document fields and the complete staged link set commit atomically on the main save.
+- Save success returns to Document Detail.
+
+## Detail Behavior
+
+- The heading shows name, type, and internal ID.
+- The header includes a contextual back action: **Back to Documents** by default, **Back to Person** when opened from Person Detail, and **Back to Job** when opened from Job Detail.
+- The detail workspace shows a Source-style pan/zoom media viewer with **Previous Page** / **Next Page** navigation for document source pages.
+- The center column is **Editable Revision** for the active source page.
+- Related People are grouped by role and link to Person Detail.
+- **Source Pages & Transcriptions** shows source/job counts and actions for source-image gallery, document jobs, and adding a Job.
+- **Edit Document**, **Print**, **Document Details**, **View Source Detail**, and **Delete** are available from the header.
+- Invalid IDs and missing Documents produce explicit states without rendering a partial page.
+
+## Document Source Images Behavior
+
+- `/documents/{document_id}/sources` shows the current Document's source pages in a thumbnail gallery.
+- Each card shows the page number, stored filename, and an **Open Source Detail** action.
+- The page includes a **Back to Document** action.
+- No source pages displays an explicit empty state.
+
+## Document Info Behavior
+
+- `/documents/{document_id}/info` contains **Archival Metadata** and **System Logistics**.
+- It includes a **Back to Document** action.
+- Archival metadata includes authors, document type, tags, document date, location (linked when present), archive identifier, and notes.
+
+## Print Behavior
+
+- Print opens a dedicated preview for persisted Document data.
+- **Facsimile** places each Source image beside its current transcription and starts every Source on a new printed sheet.
+- **Text only** omits images, joins single line breaks inside paragraphs, and preserves blank-line paragraph boundaries.
+- Non-null revised text takes precedence over raw transcription, including an intentionally empty revision.
+- Archival metadata resolves Author through the hidden built-in semantic identity, not its mutable label.
+- Archival metadata includes the Document Type label.
+- Metadata tables use a narrow non-wrapping label column and wider wrapping data columns rather than stretching across the page.
+- Job metadata uses one oldest-to-newest column per Job and ends with Status.
+- Stored text is escaped and Source media uses record-validated application URLs rather than local file paths.
+- Printing uses the browser print dialog; server-generated PDFs are not provided.
+
+## Document Jobs Behavior
+
+- The page lists the Document's Jobs newest first with status and Job ID.
+- **Open Job** navigates to Job Detail.
+- **Create Job** opens Job creation with the Document selected.
+- No jobs displays an explicit empty state.
+
+## Delete Behavior
+
+- Deletion is blocked while any Source or Job belongs to the Document.
+- The blocked state names the dependency categories and provides navigation back and to Jobs.
+- An unlinked Document requires an explicit permanent-delete action.
+- Success returns to the Documents list.
+
+## Acceptance Checklist
+
+- List columns, alignment, search, sorting, date fallback, and row navigation match this contract.
+- Create/edit enforce name, registered type, and valid exact-date input.
+- Linked People staging enforces one role and one row per Person.
+- Document and Linked People writes never partially commit.
+- Person-first Document creation preselects the requested Person as author.
+- Detail links people, Sources, and Jobs to the correct records.
+- Delete never removes a Document with Source or Job dependencies.
+- Both print formats preserve the frozen content, ordering, text-precedence, and safety contracts.
+- Service failures use the shared error presenter and never report false success.
+
+## Implementation Anchors
+
+- `src/transcription/ui/pages/documents_page.py`
+- `src/transcription/ui/components/table/documents.py`
+- `src/transcription/services/documents.py`
+- `src/transcription/services/people.py`
+- `src/transcription/services/workflows.py`
+- `src/transcription/ui/components/linked_people.py`
+- `src/transcription/ui/pages/print_preview_page.py`
+- `src/transcription/api/print_api.py`
+- `tests/ui/test_documents_page.py`
+- `tests/services/test_document_service.py`
+
+## Known Limitations and Deferred Work
+
+- Source page ordering remains read-only.
+- Printing other entities, batch printing, and server-side export formats are deferred.

docs/ui/pages/home.md

@@ -0,0 +1,64 @@
+# Home Page Contract
+
+## Purpose
+
+Home provides a user-maintained landing page for the local archive. It combines a database-backed image gallery with Markdown text and lets the operator edit both without changing application source or prompt assets.
+
+## Routes
+
+| Route | Browser path | Purpose |
+| --- | --- | --- |
+| `/homepage` | `/ui/homepage` | View homepage gallery and Markdown. |
+| `/homepage/edit` | `/ui/homepage/edit` | Upload images, manage image metadata, and edit Markdown. |
+
+The application root and `/ui` redirect to `/ui/homepage`.
+
+## View Behavior
+
+- The visible page heading is **Home**; the browser tab title is **VibeScribe Home**.
+- The featured homepage image (`photo.is_primary`) is shown first; remaining images are shown in random order.
+- The current image appears in the shared dark-room viewer with its description.
+- Saved Markdown is rendered in the **Home Text** card.
+- Missing text displays `No homepage text saved yet.`
+- Missing image displays the viewer's empty state.
+- **Edit Home Page** opens the edit route.
+- The same Home Text content is also editable from **Settings → Home Page Text**.
+
+## Edit Behavior
+
+- The image upload accepts JPEG, PNG, GIF, WebP, BMP, and TIFF files and supports multi-file uploads.
+- A successful upload immediately stores files in the shared `photo` table/media layout and displays a positive notification.
+- The editor supports per-image description edits, setting a featured image, and deleting the current image.
+- The Markdown textarea is initialized from the currently stored homepage text.
+- **Save** writes the textarea content, displays `Homepage saved`, and returns to Home.
+- **Cancel** returns to Home without saving textarea changes. An image already uploaded during the edit session remains stored.
+
+## Storage Contract
+
+- Homepage markdown text is mutable application data at `UPLOAD_DIR/homepage.md`.
+- Homepage images are stored as `photo` rows (`person_id = NULL`) with files under `UPLOAD_DIR/photos/`.
+- Uploaded images are renamed to `{photo_id}{suffix}`.
+- Homepage images are database records; markdown remains file-backed.
+
+## Acceptance Checklist
+
+- `/`, `/ui`, and the application brand reach Home.
+- Home renders with or without stored Markdown and image content.
+- Edit loads existing Markdown.
+- Supported image upload stores one or more images and makes the first image featured when no featured image exists yet.
+- Save persists Markdown and returns to Home.
+- Cancel does not save changed Markdown.
+
+## Implementation Anchors
+
+- `src/transcription/ui/pages/home_page.py`
+- `src/transcription/ui/homepage_store.py`
+- `src/transcription/ui/components/app_shell.py`
+- `tests/ui/test_upload_page.py`
+- `tests/ui/test_navigation_and_mounts.py`
+- `tests/ui/test_pages_registration.py`
+
+## Known Limitations
+
+- Homepage markdown storage location is `UPLOAD_DIR/homepage.md` and must remain writable in the active runtime environment.
+- Uploading an image is immediate and is not rolled back by Cancel.

docs/ui/pages/jobs.md

@@ -0,0 +1,99 @@
+# Jobs Page Contract
+
+## Purpose
+
+Jobs manages transcription processing runs. A Job belongs to one Document, links one or more Source pages, records processing provenance, and exposes lifecycle actions without making lifecycle fields directly editable.
+
+## Routes
+
+| Route | Purpose |
+| --- | --- |
+| `/jobs` | Searchable processing Job list. |
+| `/jobs/new` | Create and queue a Job. |
+| `/jobs/{job_id}` | View status, execution logistics, and related records. |
+| `/jobs/{job_id}/cancel` | Confirm cancellation. |
+| `/jobs/{job_id}/resubmit` | Confirm resubmission of failed Sources. |
+| `/jobs/{job_id}/delete` | Confirm or block deletion. |
+
+## List Behavior
+
+- The title is **Transcription Pipeline Jobs**.
+- **Create job** opens Job creation and **Refresh** reloads the table.
+- Columns are Job ID, Status, Document Name, # Sources, Retries, and Updated.
+- Updated is the primary date/sort field.
+- Search covers Job ID, document name, and status.
+- Status is displayed as a semantic status chip.
+- Selecting a row opens Job Detail.
+- Global Job-list row navigation includes list context so Job Detail provides **Back to Jobs**.
+- No records displays `No job records found in repository.`
+
+## Create Behavior
+
+- A Target Document and at least one source file are required.
+- `document_id` may preselect a Target Document.
+- If no Documents exist, the page explains the prerequisite and links to Document creation with a return path.
+- Provider and Model are selectable when creating a new Job.
+- Upload accepts JPEG, PNG, TIFF, and PDF files and supports multiple/folder selection.
+- The visible upload queue is sorted alphabetically by original filename.
+- Files can be removed individually or cleared before submission.
+- Helper text explains numeric filename prefixes for page ordering.
+- Submission creates the Job, Source records, and JobSource links, notifies the worker, and opens Job Detail.
+- When opened with `source_id`, creation becomes a retranscription flow: Source and Document are locked, Provider is
+  read-only, Model is restricted to `PROVIDER_MODELS`, no upload is accepted, and one existing Source is linked.
+
+## Detail and Lifecycle Behavior
+
+- The heading shows Job ID and a status badge.
+- Job Detail includes a contextual back action: **Back to Jobs** by default and **Back to Document** when opened from a Document-filtered Job list.
+- Execution Logistics shows provider, model, prompt, retry count, and last update.
+- Document Links show a clickable Document Name (with Job context), Sources count, and a single **View Sources** action using job filtering.
+- Queued and processing Jobs show an auto-refresh notice and reload every four seconds.
+- Polling stops when the Job becomes terminal or a refresh fails.
+- Queued and processing Jobs expose **Cancel**.
+- Jobs other than `transcribed` expose **Resubmit** under the current UI rule. The service blocks resubmission while processing is active or when no failed Sources exist.
+- All Jobs expose **Delete Job**, subject to explicit evidence-deletion guardrails.
+- Invalid and missing IDs produce explicit states.
+
+## Cancel Behavior
+
+- The confirmation explains that processing stops and remaining pending Sources become cancelled.
+- The service decides whether the current state permits cancellation.
+- Success updates the Job, notifies the worker, and returns to Job Detail.
+
+## Resubmit Behavior
+
+- The page shows current status and failed Source count.
+- The page explains that resubmission queues failed linked Sources while preserving immutable prior attempt evidence.
+- The service blocks submission while processing is active or when no failed Sources exist.
+- `JobSource` remains the latest compatibility projection, while every provider call appends an `ExecutionAttempt`.
+- The selected `Source.raw_transcription` projection remains available while a retry is pending or fails.
+- Success reports the number of resubmitted Sources and returns to Job Detail.
+
+## Delete Behavior
+
+- Deletion is blocked while status is `processing`.
+- Allowed deletion explicitly warns that related `JobSource` projections,
+  immutable execution attempts, captured transport responses, and attempt-owned
+  artifacts are permanently removed.
+- Source records and source files remain available for separate deletion.
+- Success returns to the Jobs list.
+
+## Acceptance Checklist
+
+- Job creation cannot proceed without a valid Document and at least one Source.
+- Upload ordering and removal controls match the displayed queue.
+- Detail shows current status and provenance summary with correct related links.
+- Active Jobs refresh without overlapping permanent polling after terminal state.
+- Cancel, resubmit, and delete honor service guardrails and show actionable failures.
+- Lifecycle fields cannot be edited directly.
+
+## Implementation Anchors
+
+- `src/transcription/ui/pages/jobs_page.py`
+- `src/transcription/ui/components/table/jobs.py`
+- `src/transcription/services/jobs.py`
+- `src/transcription/services/store.py`
+- `src/transcription/services/workflows.py`
+- `tests/ui/test_jobs_page.py`
+- `tests/services/test_job_service.py`
+- `tests/services/test_store.py`

docs/ui/pages/people.md

@@ -0,0 +1,106 @@
+# People Page Contract
+
+## Purpose
+
+People manages reusable historical-person records. A Person may appear in many Documents under different relationship roles and may optionally carry one or more photos plus a FamilySearch identifier.
+
+## Routes
+
+| Route | Purpose |
+| --- | --- |
+| `/people` | Searchable People list. |
+| `/people/new` | Create a Person. |
+| `/people/{person_id}` | View one Person and linked Documents. |
+| `/people/{person_id}/photos` | Manage Person photos. |
+| `/people/{person_id}/edit` | Edit the Person. |
+| `/people/{person_id}/delete` | Confirm permanent deletion. |
+
+## List Behavior
+
+- The title is **Archival Entities: People**.
+- **Create new person** opens the create route.
+- The table defaults to Name order (`Last Name, First & Middle`) and supports search and column sorting.
+- Columns are Last Name, First & Middle; Tags; FamilySearch ID; Birth Date; Death Date; and # Documents.
+- Name and Tags are left-aligned; FamilySearch ID, date columns, and # Documents are centered.
+- # Documents reflects how many linked Documents each Person is connected to.
+- Birth and death values independently prefer exact date, then approximate date, then `Unknown`.
+- Selecting a row opens Person Detail.
+- Row navigation includes list context so Person Detail provides **Back to People**.
+- No records displays `No person records found in repository.`
+
+## Create and Edit Behavior
+
+Required:
+
+- Last name.
+- First & middle names.
+
+Optional:
+
+- Exact and approximate birth/death dates.
+- Birth/death places.
+- Biography.
+- FamilySearch ID.
+- Tags.
+
+Rules:
+
+- Missing last name or first/middle names blocks save with a warning.
+- Exact date inputs are native browser date inputs.
+- FamilySearch IDs are normalized and validated by `PeopleService`.
+- Tags use the shared Tag registry and support inline add/select behavior.
+- Photos are managed from Person Detail via `/people/{person_id}/photos` (not in create/edit form fields).
+- Metadata JSON remains hidden.
+- Save success returns to Person Detail.
+
+## Detail Behavior
+
+- The header provides **New Document**, **Edit Person**, **Edit Photo(s)**, and **Delete**.
+- The header includes a contextual back action: **Back to People** by default, and **Back to Document** when opened from Document Detail.
+- **New Document** opens Document creation with this Person requested for author preselection.
+- Person Detail shows a single-photo viewer with **Previous/Next** navigation; the page-level **Edit Photo(s)** header action opens photo management.
+- Photo management (upload, description edit, set-primary, delete) is intentionally moved to `/people/{person_id}/photos`.
+- Biographical Record shows split names, computed full name, tags, compact birth/death dates, and places.
+- Birth and death place values are clickable links to Google Maps when present.
+- FamilySearch ID is shown as a metadata value and is clickable to the FamilySearch person details route when present.
+- Biography has an explicit empty value.
+- Linked Documents render as a table with **Document Name**, **Document Date**, **Role**, and **Number of Pages**; selecting a row opens Document Detail.
+- No links shows both an empty state and guidance to link from a Document workflow.
+- System Logistics shows created and updated timestamps.
+
+## Delete Behavior
+
+- The page warns when linked Document relationships exist.
+- Delete is blocked when related Photos exist.
+- Confirmed deletion removes the Person and its relationship links; it does not delete Documents.
+- Success returns to the People list.
+- Missing or already-deleted records return to a safe list state.
+
+## Photo Gallery Behavior (`/people/{person_id}/photos`)
+
+- Upload is triggered from a header-level **Upload Photo(s)** control beside **Back to Person**.
+- The gallery renders all photos in a responsive grid (3-4 tiles wide on larger screens).
+- Description text is shown as an overlay at the bottom of each image for quick context.
+- The editor provides **Save Description**, **Set Primary** (when applicable), and **Delete Photo** actions.
+
+## Acceptance Checklist
+
+- List fields, alignment, date fallback, search, sorting, and navigation match this contract.
+- Last name and first/middle names are enforced on create and edit.
+- FamilySearch ID validation and link generation use the fixed supported identifier format.
+- Photo upload and rendering remain constrained to supported media paths.
+- New Document carries the Person context.
+- Linked Documents show the correct role and target.
+- Delete wording distinguishes removal of relationship links from deletion of Documents.
+
+## Implementation Anchors
+
+- `src/transcription/ui/pages/people_page.py`
+- `src/transcription/ui/components/table/people.py`
+- `src/transcription/services/people.py`
+- `tests/ui/test_people_page.py`
+- `tests/services/test_v2_crud.py`
+
+## Deferred Work
+
+- Structured name fields, merge/deduplication, advanced metadata editing, and Person-side relationship editing are not current behavior.

docs/ui/pages/settings.md

@@ -0,0 +1,55 @@
+# Settings Page Contract
+
+## Purpose
+
+Settings manages installation-local registries, safe runtime .env settings, and editable text assets from one route.
+
+## Route
+
+| Route | Purpose |
+| --- | --- |
+| `/settings` | Manage Runtime Settings, Document Types, Person Roles, Tags, Prompts, Home Page Text, and Maintenance runs. |
+
+## Behavior
+
+- The page title is **Settings**.
+- Configuration surfaces are grouped as tabs:
+  - **Document Types**
+  - **Person Roles**
+  - **Tags**
+  - **Prompts**
+  - **Home Page Text**
+  - **Maintenance**
+  - **Runtime Settings**
+- Runtime Settings exposes an allowlisted set of non-secret fields synchronized with `Settings` model fields except excluded secret/unsafe fields.
+- Runtime Settings is rendered as a compact two-column editor (**Setting**, **Value**) in a centered, narrower responsive container.
+- Runtime Settings persists changes to the resolved runtime env file, validates by constructing a `Settings` instance, and reports validation failures through the shared UI error presenter.
+- `Settings` resolves its env file in this order: explicit `_env_file`, `ENV_FILE`, then the repository-root `.env.production`.
+- Runtime Settings resolves its write target in this order: explicit function override (tests/tools), `RUNTIME_SETTINGS_ENV_FILE` environment variable (deployment override), `ENV_FILE`, then the repository-root `.env.production`.
+- Runtime Settings changes require application restart to take effect.
+- Runtime Settings renders a host-side restart command (`docker compose -f docker-compose.production.yml up -d --force-recreate app worker`) so operators can apply saved values without granting Docker control to the app container.
+- Runtime Settings includes an explicit "Other settings not shown here" markdown table listing:
+  - secrets (`OPENROUTER_API_KEY`, `DATABASE__PASSWORD`)
+  - high-risk database connection settings (`DATABASE__DRIVER`, `DATABASE__PATH`, `DATABASE__HOST`, `DATABASE__PORT`, `DATABASE__DATABASE`, `DATABASE__USER`)
+    and deployment/helper keys (`CLOUDFLARE_TUNNEL_TOKEN`, `BACKUP_DIR`, `BACKUP_RETENTION_DAYS`, `RUNTIME_SETTINGS_ENV_FILE`, `ENV_FILE`, `COMPOSE_FILE`) plus legacy/deprecated keys (`POSTGRES_*`, `DATABASE_BACKUP_DIR`, `APP_DATA_BACKUP_DIR`, `UPLOADS_BACKUP_DIR`, `SYNOLOGY_BACKUP_DIR`), and directs edits for those keys to the resolved runtime env file path.
+- Document Types, Person Roles, and Tags support Add/Edit/Delete with existing guardrails.
+- Prompts exposes only `transcribe_document.md` for editing and restore-from-backup.
+- Home Page Text edits the same Markdown content rendered on `/homepage`.
+- Maintenance provides queue-backed **Run Backup** and **Run Storage Reconciliation** actions.
+- Maintenance also provides GEDCOM upload and **Run GEDCOM Import** actions, using the same queue-backed `MaintenanceRun` history/log flow.
+- Maintenance run history shows job type, status, started/finished timestamps, duration, summary, and log view/download actions.
+- Maintenance actions enqueue work and signal the worker; the page itself does not execute shell commands directly.
+
+## Acceptance Checklist
+
+- `/ui/settings` renders all seven tabs.
+- Registry and prompt workflows keep existing validation and error handling.
+- Runtime Settings excludes secret fields and rejects invalid values.
+- Saving Home Page Text persists content for the homepage view.
+
+## Implementation Anchors
+
+- `src/transcription/ui/pages/settings_page.py`
+- `src/transcription/ui/runtime_settings_store.py`
+- `src/transcription/ui/homepage_store.py`
+- `tests/ui/test_pages_registration.py`

docs/ui/pages/sources.md

@@ -0,0 +1,99 @@
+# Sources Page Contract
+
+## Purpose
+
+Sources manages individual archived page/file records. It provides source-media viewing, current processing context, provider evidence inspection, previous/next page navigation, and human revision without allowing machine output to be edited.
+
+## Routes
+
+| Route | Purpose |
+| --- | --- |
+| `/sources` | Document-filtered or Job-filtered Source list; global route redirects to Documents. |
+| `/sources/{source_id}` | View media, transcription, revision, metadata, and evidence. |
+| `/sources/{source_id}/delete` | Confirm or block deletion. |
+
+The list accepts optional `document_id` and `job_id` query parameters. Document context takes precedence if both parse successfully.
+
+## List Behavior
+
+- The global `/sources` route redirects to `/documents`.
+- Filtered list titles are **Sources for Document** and **Sources for Job**.
+- Filtered context provides **Back to Document** or **Back to Job**.
+- Rows are ordered by page number and then upload name.
+- Columns are Upload Title, Page Number, Document Name, Status, and Error Detail.
+- Document Name, Upload Title, and Error Detail are left-aligned; Status is centered.
+- Status labels are presented in uppercase for consistency with Jobs.
+- Stored Filename is intentionally absent from the list.
+- Selecting a row opens Source Detail.
+- No records displays `No source asset records found in repository.`
+
+## Detail Behavior
+
+- The heading shows page number, upload name, and Source ID.
+- **Back to Document** returns to Document Detail for the active source page.
+- **Retranscribe Source** opens Create Processing Job with this Source and its Document locked.
+- **Delete Source** opens the guarded delete route.
+- Previous and Next navigate only among Sources belonging to the same Document in page order; unavailable boundary actions are disabled.
+- The media viewer resolves the stored Source path through the configured upload root.
+- The top layout is adaptive:
+  - Standard pages use three columns with a wider Editable Revision column than the image column.
+  - Wide+narrow landscape images switch to a stacked left layout (image above Editable Revision) with metadata on the right.
+- Editable Revision is seeded from an existing revision or the preferred machine transcription.
+- Source Metadata shows upload name, stored filename, page number, Document Name, Document ID, and stored path. Source ID appears in the page-header subtitle.
+- SourceJob Metadata shows latest status (uppercase display), Job ID, execution time, provider, model, prompt, and failure detail.
+- Revision Logistics shows revised state, last-revised time, and upload time.
+- Candidate Machine Transcriptions appears below the image/revision area, remains compact until expanded, then compares it with the preferred
+  machine result and requires confirmation before **Use this transcription**.
+- Candidate promotion does not alter a human revision. Empty states distinguish no machine result from no candidates.
+- An orientation-normalized artifact appears in evidence only when recognized metadata required a physical rotation.
+
+## Provider Evidence
+
+- Provider Evidence is associated with the latest JobSource execution.
+- New attempts display separate expandable Request Manifest, Transport Response, OpenRouter SDK Response Snapshot,
+  Normalized Metadata, Software Context, and Derived Artifacts sections.
+- Historical `raw_api_response` values are labeled as OpenRouter SDK response snapshots.
+- Missing evidence has an explicit empty state.
+- Historical executions explicitly state that exact transport evidence was not captured.
+- Quality warning artifacts remain attached to their machine attempt and are not recomputed during page rendering.
+- **Export Evidence** downloads a versioned package containing source identity, attempts, artifacts, relationships,
+  schema versions, and integrity digests without source binaries, credentials, or machine-local source paths.
+
+## Revision Behavior
+
+- Machine transcription is never edited directly.
+- A revision must contain non-whitespace text.
+- Save persists revised text and updates the saved timestamp without leaving the page.
+- Reset restores the in-memory revision from page load or the most recent successful save. When no revision exists, it restores the machine transcription; it does not re-read the database.
+- A failed latest execution displays guidance that a human revision can preserve corrected text.
+
+## Delete Behavior
+
+- Deletion is allowed only when the Source has no JobSource links.
+- A linked Source shows cleanup guidance and navigation to Jobs.
+- An unlinked Source requires explicit permanent deletion.
+- Success returns to the Sources list.
+
+## Acceptance Checklist
+
+- Global, Document-filtered, and Job-filtered lists show the correct context and return action.
+- List columns and alignments match this contract and omit Stored Filename.
+- Previous/next navigation never crosses Document boundaries.
+- Detail keeps machine output read-only and human revision separately editable.
+- Retranscription, candidate comparison, warnings, and explicit promotion preserve every prior attempt.
+- Empty, failed, and missing-evidence states remain explicit.
+- JSON evidence is readable without being mislabeled as native transport evidence.
+- Delete cannot remove a Source with processing-history links.
+
+## Implementation Anchors
+
+- `src/transcription/ui/pages/sources_page.py`
+- `src/transcription/ui/components/table/sources.py`
+- `src/transcription/services/sources.py`
+- `tests/ui/test_sources_page.py`
+- `tests/services/test_transcription_service.py`
+- `tests/services/test_v2_crud.py`
+
+## Planned Changes
+
+- Source page reordering remains deferred unless a demonstrated workflow need emerges.

docs/ver1/architecture_v1.md

@@ -1,308 +0,0 @@
- # System Architecture (Version 1)
-
-This document describes the production architecture of the personal historical-document transcription system. The system is intentionally optimized for single-user operation, low operational overhead, and clean internal boundaries that support future growth without rewrites.
-
-## Architecture Objectives
-
-The production architecture is designed to:
-
-- preserve verbatim family-history source material as searchable text
-- keep operational complexity low for a personal deployment
-- support asynchronous transcription without requiring distributed infrastructure
-- maintain clear module boundaries so extensions can be added incrementally
-
-## Production Scope And Scale
-
-The deployed system targets personal use and a corpus of several thousand documents processed over time. The architecture favors simple, composable building blocks over distributed orchestration.
-
-Current scope includes:
-
-- content source upload and metadata capture
-- asynchronous transcription jobs
-- prompt-library driven transcription behavior, with one Markdown file per prompt
-- original transcription review and optional revision review
-- full-text search over accepted transcripts
-- export of transcript data
-
-## Deployment Topology
-
-The production deployment uses [Docker Compose](https://docs.docker.com/compose/) and treats containerized databases as extremely lightweight operational dependencies.
-
-Running [PostgreSQL](https://www.postgresql.org/docs/) in its own container is considered simple by default for this system.
-
-Running [MongoDB](https://www.mongodb.com/docs/) in its own container is also considered simple when document-centric storage is enabled.
-
-Container count is not a hard architectural limit; a three-container deployment (app, PostgreSQL, MongoDB) is an acceptable baseline.
-
-### Baseline Topology (Two Containers)
-
-- one application container
-- one PostgreSQL container
-- embedded background worker execution inside the app process
-
-### Expanded Topology (Three Containers)
-
-- application container
-- PostgreSQL container
-- MongoDB container
-
-No additional queue, scheduler, or search-engine containers are required in the baseline production setup.
-
-## Runtime Architecture
-
-```mermaid
-flowchart LR
-    User[Browser User] --> App[FastAPI + NiceGUI Service]
-    App --> Worker[In-process Background Worker]
-    App --> PG[(PostgreSQL)]
-    App --> MG[(MongoDB Document Store)]
-    Worker --> AI[Transcription Provider]
-    Worker --> PG
-    Worker --> MG
-```
-
-## Runtime Ownership And Startup Policy
-
-The current implementation now uses explicit lifespan-owned runtime resources.
-
-- application lifespan initializes and disposes database runtime resources
-- worker lifecycle is owned by application lifespan startup/shutdown
-- worker receives lifespan-owned database engine dependency explicitly
-- schema bootstrap policy is environment-aware and explicit:
-  - development/test default to bootstrap enabled
-  - production defaults to bootstrap disabled
-  - explicit override is available via configuration
-
-This aligns implementation toward REQ-7 and REQ-10 while preserving personal-scale operational simplicity.
-
-## Layered Module Structure
-
-### Interface Layer
-
-Responsibility:
-
-- HTTP API and UI routes
-- request/response validation
-- status and result presentation
-
-Out of scope:
-
-- business-rule enforcement
-- data-access implementation
-
-### Application Layer
-
-Responsibility:
-
-- upload and job orchestration
-- state transitions and retry policy
-- coordination across domain and infrastructure ports
-
-Out of scope:
-
-- provider-specific protocol details
-- ORM or storage-specific logic
-
-### Domain Layer
-
-Responsibility:
-
-- verbatim transcription policy
-- revision and provenance invariants
-- confidence and annotation semantics
-
-Out of scope:
-
-- web framework concerns
-- database and network I/O
-
-### Infrastructure Layer
-
-Responsibility:
-
-- persistence adapters (PostgreSQL and MongoDB)
-- transcription-provider adapter
-
-Out of scope:
-
-- business policy decisions
-
-## Processing Workflow
-
-Production transcription flow:
-
-1. A user uploads one or more content sources through the UI or API.
-2. The application validates payloads and creates document, source, and job records.
-3. The in-process worker de-queues the job and calls the transcription provider.
-4. The application persists original transcription output on the job, plus confidence metadata and provenance events.
-5. Job status transitions from queued to processing to transcribed or failed.
-6. The UI and API expose status, optional revision to original transcription, and searchable transcription text.
-
-## Data Model Ownership
-
-System-of-record entities:
-
-- documents and content sources
-- transcription jobs, original transcription, and status events
-- transcript revisions
-- provenance metadata
-
-### Original Transcription And Revision Ownership
-
-- each processing job stores the original immutable provider output (`text`)
-- provider metadata (`provider`, `model`, `prompt_name`) and failure detail (`error_detail`) are job-owned processing artifacts
-- revisions are optional user-authored edits linked to a content source
-- a revision can be created from original `job.text`
-- many jobs will have zero revisions; revisions are additive and never overwrite original provider output
-- a document groups one or more content sources (images, PDFs, and future source types)
-
-Storage strategy:
-
-- PostgreSQL for relational system-of-record entities
-- MongoDB for document-oriented payloads and large transcription artifacts
-- versioned prompt artifacts stored as individual Markdown files for human editing and refinement
-- in-memory execution state treated as ephemeral
-
-## Transcription Prompt Asset Policy
-
-The production system treats transcription prompts as maintainable content assets.
-
-- each transcription prompt is stored in its own Markdown file
-- prompt files are designed for direct human editing and iterative refinement
-- prompt updates are independent and do not require bundling unrelated prompt changes
-- prompt file identity and revision history are tracked through normal repository version control
-
-## Simplicity Guardrails
-
-The production system enforces these constraints to prevent accidental over-engineering:
-
-- PostgreSQL in a container is treated as a lightweight default dependency
-- MongoDB in a container is treated as a lightweight optional dependency
-- three containers (app, PostgreSQL, MongoDB) is an acceptable simple deployment
-- no dedicated queue or search cluster is introduced without measured need
-- external infrastructure is added only behind existing ports/adapters
-
-## Extension Path
-
-The architecture supports additive growth without changing domain contracts.
-
-### Stage 1: Foundation (Current)
-
-- upload, transcription, review, search, export
-- in-process worker execution
-- single provider adapter
-- app plus PostgreSQL deployment
-
-### Stage 2: Throughput Hardening
-
-- optional MongoDB document-store enablement
-- optional external worker/queue process
-- stronger retry and dead-letter handling
-
-### Stage 3: Intelligence Features
-
-- entity extraction and cross-document linking
-- timeline and narrative assembly
-- optional multi-provider routing
-
-Each stage preserves existing module boundaries and keeps migration risk low.
-
-## Test Strategy
-
-The test strategy is aligned to personal-scale operation with fast, deterministic feedback.
-
-### Unit Tests
-
-- domain transcription rules and annotation behavior
-- revision-history invariants
-- job state-transition logic
-
-### Integration Tests
-
-- repository behavior and transaction boundaries
-- persistence-adapter and provider adapter contract mapping
-- upload-to-persistence roundtrip
-
-### End-to-End Tests
-
-- happy path: upload, transcribe, review, search, export
-- failure path: provider error, retry, surfaced failed status
-
-### CI Execution Model
-
-- fast suite on each push
-- optional slower provider-sandbox checks on scheduled runs
-
-## Risks And Controls
-
-### Runtime Responsiveness
-
-Risk:
-
-- long jobs can reduce responsiveness in a single-process deployment
-
-Control:
-
-- bounded concurrency and visible job status in the UI
-
-### Database Concurrency Limits
-
-Risk:
-
-- contention can appear under sustained concurrent writes in personal-scale infrastructure
-
-Control:
-
-- tuned connection pooling and phased use of MongoDB for document-heavy workloads
-
-### Provider Output Variance
-
-Risk:
-
-- transcription quality varies by content source type, handwriting legibility, and source quality
-
-Control:
-
-- first-class human review and immutable revision history
-
----
-
-## Technology References
-
-- [FastAPI documentation](https://fastapi.tiangolo.com/)
-- [NiceGUI documentation](https://nicegui.io/documentation)
-- [Docker Compose documentation](https://docs.docker.com/compose/)
-- [PostgreSQL documentation](https://www.postgresql.org/docs/)
-- [MongoDB documentation](https://www.mongodb.com/docs/)
-
-## Related Local References
-
-- [System Overview](index_v1.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- System Architecture (this document)
-- [System Requirements](requirements_v1.md)
-- [Data model](schema_v1.md)
-- [Error Handling Policy](error_handling_v1.md)
-- [Implementation Plan](implementation_plan_v1.md)
-
-## Glossary
-
-- Adapter: A component that translates between internal interfaces and external systems such as databases or AI services.
-- Background job: Work executed outside the request/response path so the UI remains responsive.
-- Boundary: A strict separation between modules with different responsibilities.
-- CI (Continuous Integration): Automated test execution for code changes.
-- Contract test: A test that verifies an adapter follows expected input/output behavior at a boundary.
-- Domain layer: The module that contains core business rules and invariants.
-- End-to-end test: A test that validates a full user flow across the running system.
-- Full-text search: Text indexing and querying optimized for natural-language search.
-- In-process worker: A background executor that runs within the same application process.
-- Integration test: A test that verifies interactions between real modules and infrastructure components.
-- MongoDB: A document-oriented database used for flexible, high-variance data structures.
-- Modular monolith: A single deployable application with strongly separated internal modules.
-- Port/Interface: A stable contract used by application/domain code to call infrastructure implementations.
-- Prompt artifact: A single Markdown file that defines one transcription prompt and can be revised independently.
-- Provenance: Metadata that records where generated data came from and how it was produced.
-- Revision history: Optional versioned record of user-authored transcription edits over time.
-- System of record: The authoritative persistent store for canonical data.
-- Vertical slice: A minimal end-to-end feature path spanning UI/API, application logic, and persistence.

docs/ver1/error_handling_v1.md

@@ -1,288 +0,0 @@
-# Error Handling Policy
-
-This document defines the canonical error-handling policy for the document transcription system. It is the single source of truth for how errors are classified, surfaced to users, logged for diagnosis, and handled across UI, API, service, worker, and provider boundaries.
-
-## Error Handling Objectives
-
-The production error-handling model is designed to:
-
-- make failures visible to the user in clear, actionable language
-- preserve enough diagnostic detail for fast troubleshooting
-- keep module behavior consistent across all boundaries
-- distinguish expected domain failures from unexpected defects
-- support safe retries for transient failures without hiding persistent faults
-
-## Scope And Authority
-
-This page governs error-handling behavior for:
-
-- UI interactions (NiceGUI pages)
-- API endpoints (FastAPI routes)
-- application services and orchestration logic
-- in-process background worker execution
-- external provider adapters and persistence adapters
-
-If implementation behavior conflicts with this document, this document is authoritative and implementation should be updated.
-
-## Core Principles
-
-- **Clarity first:** user-facing messages should explain what failed in plain language.
-- **Actionability required:** each surfaced error should include a suggested next step.
-- **Safety by default:** internal details are logged; sensitive details are not exposed by default in UI/API.
-- **Consistency across boundaries:** category and structure should remain stable from source to surface.
-- **Fail explicitly:** silent failure is prohibited.
-- **Traceability:** every non-trivial error should be traceable with an error reference ID.
-
-## Error Taxonomy
-
-The system uses stable, implementation-independent categories:
-
-| Category | Definition | Typical Source | Retriable |
-| --- | --- | --- | --- |
-| `validation_error` | Payload or parameter shape/content is invalid | UI/API input validation, service guards | no |
-| `user_input_error` | User-provided artifact is unacceptable though structurally valid | unsupported file type, empty file, oversized upload | sometimes |
-| `not_found_error` | Requested resource does not exist | missing job/document/source/revision | no |
-| `conflict_error` | Requested operation violates current state constraints | invalid state transition | no |
-| `external_provider_error` | External AI/provider call fails | upstream HTTP/API/provider failures | sometimes |
-| `infrastructure_transient_error` | Temporary environment issue | network timeout, DB connection reset | yes |
-| `infrastructure_persistent_error` | Non-transient environment issue | missing permissions, misconfiguration | no |
-| `internal_unexpected_error` | Unhandled defect or unknown failure | uncaught exceptions, logic errors | unknown (default no) |
-
-### Classification Rules
-
-- Classification occurs as close as possible to the origin boundary.
-- Provider-specific exceptions must be normalized into taxonomy categories before crossing service boundaries.
-- Unknown exceptions are classified as `internal_unexpected_error` and logged with traceback.
-- Category names are stable contracts and must not be changed casually.
-
-## User-Facing Error Experience Contract
-
-When an error is shown in the GUI, it must include:
-
-1. **Title** (short context, e.g., “Upload failed”)
-2. **Message** (plain-language explanation)
-3. **Suggested action** (explicit next step)
-4. **Error reference ID** (for support/debug traceability)
-5. **Technical details** (optional/collapsible for advanced users)
-
-### UI Message Rules
-
-- Do not expose raw stack traces by default.
-- Do not expose secrets, credentials, connection strings, or filesystem internals unless explicitly in debug tooling.
-- Prefer domain language over implementation language.
-- Use persistent visibility for important failures (dialog/card), not only transient toasts.
-
-### Suggested Action Requirements
-
-Every user-visible error must include a suggested course of action, such as:
-
-- retry the operation
-- check file type/size constraints
-- refresh the jobs page
-- verify environment configuration
-- contact operator with error ID and timestamp
-
-## API Error Response Contract
-
-API errors should return a structured envelope with stable fields:
-
-- `error_id`: short unique reference ID
-- `category`: taxonomy category
-- `message`: safe human-readable summary
-- `suggestion`: recommended next step
-- `details`: optional, only when safe and appropriate
-- `timestamp`: UTC ISO-8601
-
-HTTP status mapping guidance:
-
-- `validation_error`, `user_input_error` -> `400`
-- `not_found_error` -> `404`
-- `conflict_error` -> `409`
-- `external_provider_error` -> `502` or `503` (depending on failure mode)
-- `infrastructure_transient_error` -> `503`
-- `infrastructure_persistent_error` -> `500`
-- `internal_unexpected_error` -> `500`
-
-## Logging And Observability Contract
-
-All logged errors must include, where available:
-
-- `error_id`
-- `category`
-- `operation` (e.g., `upload.submit`, `worker.process_job`, `jobs.refresh`)
-- `exception_type`
-- `job_id`, `document_id`, `source_id` (when relevant)
-- UTC timestamp
-
-Rules:
-
-- Use structured logging fields where practical.
-- Use full traceback for unexpected errors (`internal_unexpected_error`).
-- Log at boundary handoff points to preserve causal trail.
-- Avoid duplicate noisy logging for the same exception at every layer.
-
-## Recovery And Retry Policy
-
-### Retriable Conditions
-
-Retriable failures include:
-
-- transient network/provider timeouts
-- intermittent provider unavailability
-- temporary DB/network interruptions
-
-### Non-Retriable Conditions
-
-Non-retriable failures include:
-
-- invalid file formats
-- missing required data
-- permission/configuration failures
-- deterministic domain conflicts
-
-### Worker Behavior
-
-- The worker must classify and persist failure details consistently.
-- Retries should be bounded by configured limits.
-- Exhausted retries must end in explicit failed status with recorded reason.
-- No infinite retry loops are allowed.
-
-## Boundary-Specific Responsibilities
-
-### UI Layer
-
-Responsibility:
-
-- display user-safe error summaries and suggested actions
-- show persistent error visibility for critical failures
-- include error reference IDs in visible output
-
-Out of scope:
-
-- low-level exception parsing
-- provider-specific protocol interpretation
-
-### API Layer
-
-Responsibility:
-
-- map application exceptions into stable error envelopes and HTTP statuses
-- preserve category and error_id continuity
-
-Out of scope:
-
-- domain-specific remediation logic
-
-### Service Layer
-
-Responsibility:
-
-- classify domain and infrastructure exceptions
-- convert adapter-specific failures into taxonomy categories
-- return deterministic error types to callers
-
-Out of scope:
-
-- presentation formatting for UI
-
-### Worker Layer
-
-Responsibility:
-
-- execute retry policy for retriable failures
-- persist terminal failure details for jobs
-- emit operational logs with category and identifiers
-
-Out of scope:
-
-- direct UI messaging
-
-### Provider Adapter Layer
-
-Responsibility:
-
-- normalize provider SDK/HTTP failures into domain-neutral exceptions
-- preserve raw provider context for logs (safely)
-
-Out of scope:
-
-- choosing user-facing wording
-
-## Error Lifecycle Workflow
-
-Standard lifecycle:
-
-1. Failure occurs at a boundary or operation.
-2. Exception is classified into taxonomy category.
-3. `error_id` is created (or propagated).
-4. Error is logged with required structured fields.
-5. User/API receives safe message + suggested action.
-6. Persistent job/resource state is updated when applicable.
-7. Tests verify contract behavior for the pathway.
-
-## Test Strategy For Error Handling
-
-### Unit Tests
-
-- category classification behavior
-- retry eligibility decisions
-- exception-to-message mapping safety
-
-### Integration Tests
-
-- UI pathways show clear message + suggested action for known failures
-- API returns structured error envelope with expected status/category
-- worker persists failed status and failure detail as required
-
-### Regression Tests
-
-- each previously observed production issue should have a guarding test
-- contract tests must cover adapter error normalization behavior
-
-## Known Failure Patterns And Prescribed Responses
-
-| Pattern | Category | User Message | Suggested Action |
-| --- | --- | --- | --- |
-| Upload payload cannot be parsed by UI handler | `internal_unexpected_error` (until narrowed) | Upload failed due to unexpected processing error | Retry once; if repeated, report error ID and check runtime version compatibility |
-| Unsupported extension | `user_input_error` | File type is not supported | Upload JPG, PNG, TIFF, or PDF |
-| Empty file upload | `validation_error` | Uploaded file is empty | Choose a valid non-empty file and retry |
-| Provider timeout | `external_provider_error` or `infrastructure_transient_error` | Transcription provider timed out | Retry from jobs page; if repeated, check provider status |
-| Job lookup missing | `not_found_error` | Requested job was not found | Refresh jobs list and open a valid job |
-
-## Governance And Update Process
-
-This document is a living policy artifact.
-
-Update this document when:
-
-- new error categories are introduced
-- handling behavior changes at any boundary
-- a production incident reveals missing guidance
-- API/UI error contracts change
-
-Change requirements:
-
-- update this document and associated tests in the same change set
-- preserve taxonomy stability; if changed, document migration impact
-- record noteworthy policy changes in project release notes or changelog
-
----
-
-## Related Local References
-
-- [System Overview](index_v1.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v1.md)
-- [System Requirements](requirements_v1.md)
-- [Data model](schema_v1.md)
-- Error Handling Policy (this document)
-- [Implementation Plan](implementation_plan_v1.md)
-
-## Glossary
-
-- Error category: Stable classification used to drive handling, messaging, and status mapping.
-- Error envelope: Structured API payload describing a failure.
-- Error reference ID: Short identifier used to correlate user-visible failure with logs.
-- Retriable error: Failure likely to succeed on a later attempt without code changes.
-- Terminal failure: Failure state after retries are exhausted or retry is not allowed.

docs/ver1/implementation_plan_v1.md

@@ -1,203 +0,0 @@
-# Version 1 Implementation Plan
-
-This plan defines the path from current implementation to **Version 1 complete**, aligned to the updated domain model:
-
-- `Document` groups one or more content `Source` records
-- `Job` owns original immutable provider output (`text`) and processing metadata
-- `Revision` stores optional user-authored edits linked to a `Source`
-
-The objective is to complete V1 scope with production readiness while keeping non-V1 enhancements out of active delivery.
-
----
-
-## V1 Completion Definition
-
-V1 is complete when all of the following are true:
-
-1. **Functional complete**
-   - Upload, queue, processing, status display, and transcription result inspection work end-to-end.
-   - Optional revision workflow is implemented (create/view/update single revision).
-2. **Data-model complete**
-   - Runtime behavior, persistence, and tests all align to `Document` / `Source` / `Job` / `Revision`.
-3. **Operational complete**
-   - Error handling, logs, and runbooks support reliable operation.
-4. **Documentation complete**
-   - Architecture, requirements, schema, error handling, and index are consistent and current.
-
----
-
-## Phase 1 — Data Contract Stabilization (Schema-First)
-
-**Goal:** Lock a single canonical contract before further feature work.
-
-### Tasks
-1. Confirm and document invariants:
-   - `Job.text` is original immutable transcription output.
-   - `Revision` is optional and user-authored.
-   - Revisions are derived from the original `Job.text`.
-2. Verify relationship cardinality assumptions:
-   - `Document` -> many `Source`
-   - `Document` -> many `Job`
-   - `Source` -> one `Job`
-   - `Source` -> one `Revision`
-3. Ensure field naming consistency (`date_created`, `date_updated`, `date_uploaded`) across code and docs.
-4. Freeze V1 status lifecycle to current implementation (`queued`, `processing`, `transcribed`, `failed`).
-
-### Deliverables
-- Updated `schema_v1.md` and `requirements.md` traceability alignment.
-- Explicit V1 data invariants section in architecture docs.
-
-### Exit Criteria
-- No conflicting definitions of ownership/cardinality/status remain in docs.
-
----
-
-## Phase 2 — Service Layer Refactor To New Model
-
-**Goal:** Remove all obsolete `Transcript` assumptions from service/workflow code.
-
-### Tasks
-1. Refactor `services/transcription.py`:
-   - Replace transcript CRUD assumptions with job-output + revision operations.
-2. Refactor `services/jobs.py`:
-   - Replace old timestamp/relationship accessors with current model fields.
-3. Refactor `services/documents.py` and `services/store.py`:
-   - Ensure upload creates and links `Document`, `Source`, and `Job` correctly.
-4. Refactor `services/workflows.py`:
-   - Persist original provider output to `Job`.
-   - Persist failure detail to `Job.error_detail`.
-   - Use `Revision` only for user-authored edits.
-
-### Deliverables
-- Service layer fully aligned with new schema.
-
-### Exit Criteria
-- No service module imports or persists `Transcript` model artifacts.
-
----
-
-## Phase 3 — UI Contract Alignment
-
-**Goal:** Align pages/components to source/job/revision semantics.
-
-### Tasks
-1. Update job detail and related UI components:
-   - Display original immutable transcription from `Job.text`.
-   - Display optional revision sourced from `Source.revision` (0 or 1).
-2. Align date fields with new schema naming.
-3. Preserve clear user messaging when no revisions exist.
-
-### Deliverables
-- Updated jobs page and detail components.
-
-### Exit Criteria
-- UI behavior and labels match documentation and domain model.
-
----
-
-## Phase 4 — Database Bootstrap, Migration, and Safety
-
-**Goal:** Make schema transition safe in dev/test and repeatable for deployment.
-
-### Tasks
-1. Update bootstrap compatibility logic in `db/operations.py`:
-   - Remove obsolete transcript-table assumptions.
-   - Add forward-compatible patches for current tables only.
-2. Define migration/backfill approach for existing local data.
-3. Document rollback and recovery steps.
-4. Rehearse migration path against representative data.
-
-### Deliverables
-- Migration/upgrade runbook.
-- Validated bootstrap behavior for dev/test.
-
-### Exit Criteria
-- Migration path is documented and tested with no unresolved data-loss risk.
-
----
-
-## Phase 5 — Test Suite Realignment
-
-**Goal:** Restore full confidence after the schema redesign.
-
-### Tasks
-1. Rewrite model tests for:
-   - `Document`, `Source`, `Job`, `Revision` relationships and invariants.
-2. Rewrite service/integration tests:
-   - Worker success/failure paths using `Job.text` / `Job.error_detail`.
-   - Optional single-revision creation/update behavior.
-3. Update UI tests for new job-detail/revision rendering behavior.
-4. Re-enable strict CI quality gates (lint, type, tests).
-
-### Deliverables
-- Updated test matrix and passing CI.
-
-### Exit Criteria
-- Critical user flows and failure paths are covered and green.
-
----
-
-## Phase 6 — Reliability, Operations, and Release Readiness
-
-**Goal:** Ensure V1 is operable and launch-safe.
-
-### Tasks
-1. Verify error taxonomy behavior across UI/API/service/worker.
-2. Confirm structured logging includes relevant identifiers (`job_id`, `document_id`, `source_id` when applicable).
-3. Validate retry behavior and terminal failure handling.
-4. Finalize release checklist, deployment steps, and rollback procedure.
-5. Execute final acceptance run against requirements traceability.
-
-### Deliverables
-- V1 release checklist and acceptance evidence.
-- `runbook_v1.md` for incident response and operator workflows.
-- `release_checklist_v1.md` for release sign-off.
-
-### Exit Criteria
-- Stakeholder sign-off and launch readiness achieved.
-
----
-
-## Requirement Traceability Focus
-
-The plan must keep clear evidence against these requirement groups:
-
-- **Core flow:** REQ-0 to REQ-6
-- **Runtime and operations constraints:** REQ-7 to REQ-12
-- **Revision workflow:** REQ-13
-
-A lightweight traceability table should be maintained with:
-
-- requirement ID
-- implementation status (`not started` / `in progress` / `done`)
-- validation evidence (test name, screenshot, or runbook step)
-
----
-
-## Suggested Execution Rhythm
-
-- **Weekly:** requirement status and risk review
-- **Per PR:** contract checks (model names, field names, lifecycle values)
-- **Milestone checks:** end of Phases 2, 4, and 6
-
----
-
-## Scope Discipline Rule (V1 Focus)
-
-- Only work required to satisfy V1 requirements enters this plan.
-- Nice-to-have enhancements are captured in a separate backlog document.
-- Schema or contract changes after Phase 1 require explicit approval and traceability impact review.
-
----
-
-## Related Local References
-
-- [System Overview](index_v1.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v1.md)
-- [System Requirements](requirements_v1.md)
-- [Data model](schema_v1.md)
-- [Error Handling Policy](error_handling_v1.md)
-- Implementation Plan (this document)
-

docs/ver1/index_v1.md

@@ -1,58 +0,0 @@
-## Document Transcription System Overview
-
-This project is a production application for transcribing and preserving historical family documents. It is intentionally designed for personal-scale use, with a simplicity-first architecture that is easy to operate and easy to extend.
-
-## Start Here
-
-Read [architecture_v1.md](architecture_v1.md) first.
-
-The architecture page is the primary technical reference and defines:
-
-- deployed topology and infrastructure limits
-- module boundaries and dependency flow
-- processing life cycle and data ownership
-- test strategy, risk controls, and extension path
-
-## What The Application Does
-
-At a high level, users upload images or PDFs as content sources for handwritten, typed, or typeset documents, run asynchronous transcription jobs, review optional revisions, and search across accepted text.
-
-### Core capabilities:
-
-- document grouping with one or more content sources and metadata capture
-- asynchronous transcription with visible job status
-- immutable original transcription persisted with each job (plus provider/model/prompt metadata)
-- transcription prompt management with one Markdown file per prompt for human refinement over time
-- optional revisions for user-authored edits of original immutable transcription text
-- full-text search over accepted transcripts
-- export of transcript data
-
-## Production Operating Model
-
-The system runs with minimal operational overhead:
-
-- PostgreSQL in a dedicated Docker container is considered extremely lightweight and simple for this system
-- MongoDB in a dedicated Docker container is also considered extremely lightweight and simple for document-centric persistence
-- a three-container deployment (app, PostgreSQL, MongoDB) is a simple and acceptable baseline
-- no required queue or search-engine containers in the baseline setup
-
-This operating model keeps deployment and maintenance simple while preserving clean boundaries for future scale.
-
----
-
-## Documentation Map
-
-- System Overview (this document)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v1.md)
-- [System Requirements](requirements_v1.md)
-- [Data model](schema_v1.md)
-- [Error Handling Policy](error_handling_v1.md)
-- [Implementation Plan](implementation_plan_v1.md)
-
-## Glossary
-
-- Document-oriented persistence: Storing data as flexible records instead of fixed relational rows.
-- Prompt artifact: A single Markdown file that defines one transcription prompt and is edited independently.
-- System of record: The authoritative persistent store for canonical data.

docs/ver1/release_checklist_v1.md

@@ -1,45 +0,0 @@
-# V1 Release Readiness Checklist
-
-Use this checklist before declaring V1 operationally complete.
-
-## A) Functional Readiness
-
-- [ ] Upload flow works for supported file types.
-- [ ] Worker transitions jobs through `queued -> processing -> transcribed|failed`.
-- [ ] Job detail displays immutable original transcription from `Job.text`.
-- [ ] Revision workflow supports create/update/view/delete for optional single revision.
-
-## B) Reliability and Error Handling
-
-- [ ] Error categories surface with actionable messages in UI/API pathways.
-- [ ] Failed jobs persist `error_detail` and terminal state.
-- [ ] Stale processing recovery verified on restart.
-- [ ] Retry/timeout behavior validated against configured limits.
-
-## C) Operational Readiness
-
-- [ ] `runbook_v1.md` reviewed and current.
-- [ ] `migration_v1.md` reviewed and current.
-- [ ] Backup and rollback procedures tested at least once.
-- [ ] Incident escalation packet template is known to operators.
-
-## D) Quality Gates
-
-- [ ] Lint/type checks pass.
-- [ ] `pytest -m "not external" -q` passes.
-- [ ] Targeted external/provider checks executed (if credentials available).
-- [ ] Release evidence recorded in `release_evidence_v1.md`.
-
-## E) Traceability and Documentation
-
-- [ ] `requirements_v1.md` aligns with implemented V1 behavior.
-- [ ] `architecture_v1.md`, `schema_v1.md`, and `error_handling_v1.md` are consistent.
-- [ ] `traceability_v1.md` is updated with current implementation and test evidence.
-- [ ] `implementation_plan_v1.md` phase status updated with evidence references.
-- [ ] REQ traceability evidence links recorded (tests/runbook/checks).
-
-## Release Sign-Off
-
-- [ ] Technical sign-off complete.
-- [ ] Operational sign-off complete.
-- [ ] V1 completion date recorded.

docs/ver1/release_evidence_v1.md

@@ -1,39 +0,0 @@
-# V1 Release Evidence Log
-
-## Step 5 Quality Gates (2026-07-29)
-
-### Lint
-
-- Command: `python -m ruff check .`
-- Result: ✅ pass
-- Notes: initial findings were auto-fixed (`ruff --fix`) plus small manual line-wrap/annotation adjustments.
-
-### Tests (primary gate)
-
-- Command: `python -m pytest -m "not external" -q`
-- Result: ✅ pass (`[100%]`)
-
-### Tests (external smoke)
-
-- Command: `python -m pytest -m external -q`
-- Result: ✅ pass (`[100%]`)
-
-### Type Check
-
-- Command: `python -m ty check src tests`
-- Result: ⚠️ not passing
-- Summary: existing SQLModel/SQLAlchemy typing incompatibilities and test double typing mismatches remain.
-
-Key current blocker families:
-
-1. SQLModel relationship/query attribute typing (`selectinload`, `order_by`, `.any()`)
-2. SQLAlchemy join clause typing in `services/transcription.py`
-3. Test fake client type mismatch for `OpenRouterTranscriptionProvider(client=...)`
-4. `Settings(**defaults)` typed-dict strictness in `tests/test_config.py`
-
-## Current Gate Status
-
-- Lint: pass
-- Non-external tests: pass
-- External smoke tests: pass
-- Type check: **blocked** (requires dedicated typing cleanup pass)

docs/ver1/requirements_v1.md

@@ -1,98 +0,0 @@
-## Document Transcription System Requirements
-
-This page captures a SysML v1.6-style requirements baseline for the production system described in [index_v1.md](index_v1.md). The model is represented as concise tables and traceability lists that preserve SysML-style IDs and relationship semantics.
-
-## Scope
-
-- System of interest: the single Python application service (NiceGUI + FastAPI) with PostgreSQL as the relational system of record and optional MongoDB for document-oriented persistence.
-- Operational context: local-first execution with Docker Compose and an intentionally lightweight production trajectory.
-- Primary concern: end-to-end transcription job lifecycle from upload through completion or failure.
-
-## Requirements Model (Concise Text Form)
-
-### Requirements
-
-| ID | Category | Requirement | Risk | Verify Method |
-| --- | --- | --- | --- | --- |
-| REQ-0 | System | Provide end-to-end document transcription with persistent, inspectable lifecycle state. | medium | demonstration |
-| REQ-1 | Functional | Allow users to upload one or more images or PDFs as sources from the web UI. | low | test |
-| REQ-2 | Functional | Run each upload through asynchronous processing that returns an original transcription or explicit failure. | high | test |
-| REQ-3 | Functional | Persist and expose job states: queued, processing, transcribed, failed. | high | inspection |
-| REQ-4 | Functional | Persist transcription output, processing history, and failure details. | medium | test |
-| REQ-5 | Interface | Expose API and UI views for status inspection and completed transcription reading. | medium | demonstration |
-| REQ-6 | Performance | Trigger background processing on upload to preserve UI responsiveness. | medium | analysis |
-| REQ-7 | Design Constraint | Keep lifespan-owned runtime resources: SQLAlchemy engine, async session factory, worker resources, provider clients. | medium | inspection |
-| REQ-8 | Design Constraint | Initialize configuration and logging once at startup through centralized mechanisms. | low | inspection |
-| REQ-9 | Design Constraint | Use Docker Compose baseline of app plus PostgreSQL; allow optional MongoDB container when enabled. | medium | demonstration |
-| REQ-10 | Design Constraint | Keep schema bootstrap explicit and opt-in; normal startup does not mutate production schema. | high | inspection |
-| REQ-11 | Design Constraint | Use service-backed persistence for core document and job data. | medium | inspection |
-| REQ-12 | Design Constraint | Store transcription prompts as individual Markdown artifacts for iterative refinement. | medium | inspection |
-| REQ-13 | Functional | Allow users to create one optional revision of transcription text derived from the original job transcription. | low | test |
-
-### Requirement Relationships
-
-- Contains: REQ-0 contains REQ-1 through REQ-13.
-- Derives: REQ-2 -> REQ-3, REQ-3 -> REQ-4.
-- Traces: REQ-5 -> REQ-3.
-- Refines: REQ-6 -> REQ-2.
-
-### Architecture Elements
-
-| Element | Type | Doc Reference |
-| --- | --- | --- |
-| UI | NiceGUI pages | src/transcription/ui/pages |
-| API | FastAPI routes | src/transcription/api/routes.py |
-| GRAPH | Async processing workflow | src/transcription/services, src/transcription/ai |
-| DBREL | PostgreSQL + SQLModel relational persistence | src/transcription/db |
-| DBDOC | MongoDB document persistence | src/transcription/db, src/transcription/services |
-| OPS | Docker Compose runtime | docker-compose.yml |
-| PROMPTS | Transcription prompt artifact library (Markdown files) | .github/prompts, docs |
-| TESTS | Pytest verification suite | tests |
-
-### Satisfaction Mapping
-
-- UI satisfies REQ-1, REQ-5, REQ-13.
-- API satisfies REQ-5.
-- GRAPH satisfies REQ-2, REQ-6.
-- DBREL satisfies REQ-3, REQ-10, REQ-13.
-- DBDOC satisfies REQ-4, REQ-11.
-- OPS satisfies REQ-9.
-- PROMPTS satisfies REQ-12.
-
-### Verification Mapping
-
-- TESTS verifies REQ-1, REQ-2, REQ-3, REQ-4, REQ-5, REQ-10, REQ-11, REQ-12, REQ-13.
-
-## Requirement Notes
-
-- Requirement IDs (`REQ-*`) are stable references for planning, implementation, and test traceability.
-- The model uses compact tables and traceability lists for renderer compatibility while preserving SysML-style requirement IDs and relationship semantics.
-- Requirement categories (functional, interface, performance, and design constraints) are preserved as explicit REQ entries and relationship labels to keep change impact visible.
-- PostgreSQL containerization and optional MongoDB containerization are both treated as extremely lightweight and simple operational choices in this architecture.
-
-## Verification Intent
-
-- Demonstration: validate end-to-end behavior via running system flows and operator-visible outcomes.
-- Inspection: verify architecture and startup/runtime policies in code and configuration.
-- Analysis: evaluate asynchronous execution behavior and design sufficiency.
-- Test: automate behavioral checks through pytest suites and service-level tests.
-
----
-
-## Related Local References
-
-- [System Overview](index_v1.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v1.md)
-- System Requirements (this document)
-- [Data model](schema_v1.md)
-- [Error Handling Policy](error_handling_v1.md)
-- [Implementation Plan](implementation_plan_v1.md)
-
-## Glossary
-
-- Document-oriented persistence: A storage approach that uses flexible document structures for variable data shapes.
-- Prompt artifact: A single Markdown file that defines one transcription prompt and is revised independently.
-- SysML: Systems Modeling Language used to express structured requirements and traceability.
-- System of record: The authoritative persistent store for canonical business data.

docs/ver1/runbook_v1.md

@@ -1,129 +0,0 @@
-# V1 Operations Runbook
-
-This runbook provides day-2 operational procedures for the V1 baseline.
-
-## Scope
-
-Applies to:
-
-- local/hosted V1 runtime
-- SQLite-backed persistence
-- in-process worker lifecycle
-- OpenRouter provider integration
-
-## Preconditions
-
-- `.env` contains `OPENROUTER_API_KEY`
-- app starts successfully
-- `uploads/` and `prompts/` are writable
-- health endpoint responds at `/healthz`
-
-## Standard Startup Procedure
-
-1. Start the app using the project-standard command.
-2. Open `/healthz` and verify `{"status":"ok"}`.
-3. Open `/ui/upload` and submit a small valid file.
-4. Confirm job transitions from `queued` -> `processing` -> `transcribed` (or `failed` with detail).
-
-## Standard Shutdown Procedure
-
-1. Stop the application process.
-2. Ensure no active process still holds the SQLite file.
-3. If maintenance is planned, copy the DB file before edits:
-   - `transcription.db` (or configured `DATABASE_URL` file path)
-
-## Incident: Jobs Stuck In `processing`
-
-### Symptoms
-
-- Jobs remain `processing` for longer than provider timeout
-- New uploads queue but do not complete
-- provider usage increases but no terminal job state is visible
-
-### Checks
-
-1. Confirm app process is still running.
-2. Confirm worker loop is active (startup logs include worker lifespan start).
-3. Inspect recent app logs for:
-   - `worker.process_job`
-   - `error_id`
-   - `category`
-   - `job_id` / `document_id` / `source_id`
-4. Verify provider credentials and provider status.
-
-### Recovery
-
-1. Restart the app to trigger stale-processing recovery.
-2. On startup, app re-queues stale processing jobs based on timeout policy.
-3. Re-check jobs page and confirm terminal state progression.
-4. If persistent, capture logs + error IDs and move to deep investigation.
-
-## Incident: Provider Authentication Failures
-
-### Symptoms
-
-- failures categorized as provider/auth
-- jobs fail quickly with authentication guidance
-
-### Recovery
-
-1. Validate `OPENROUTER_API_KEY` value.
-2. Restart app after updating env.
-3. Re-run a small transcription to confirm recovery.
-
-## Incident: Upload Failures
-
-### Symptoms
-
-- UI reports upload errors
-- unsupported extension or empty payload
-
-### Recovery
-
-1. Validate file extension (`.jpg`, `.jpeg`, `.png`, `.tif`, `.tiff`, `.pdf`).
-2. Validate file is not empty.
-3. Validate upload directory permissions.
-4. Retry upload.
-
-## Incident: Database File/Permission Issues
-
-### Symptoms
-
-- persistence errors during upload/job update
-- startup failures around schema/runtime
-
-### Recovery
-
-1. Confirm the configured DB file path exists and is writable.
-2. Confirm parent directory permissions.
-3. Restore from last known backup copy if corruption is suspected.
-4. Restart app and run smoke test.
-
-## Logging Requirements (Operational)
-
-Operational triage should always capture:
-
-- `error_id`
-- category
-- operation name
-- `job_id`, `document_id`, `source_id` when applicable
-- UTC timestamp
-
-## Escalation Packet (When opening an issue)
-
-Include:
-
-- exact timestamp window
-- one failing `job_id`
-- relevant `error_id` values
-- latest 100 lines of app logs
-- environment summary (`DATABASE_URL` type, app version/commit)
-
-## Post-Incident Validation
-
-After mitigation, verify:
-
-1. Upload works.
-2. One job reaches `transcribed`.
-3. One induced failure reaches `failed` with error detail.
-4. Jobs page and detail page render correctly.

docs/ver1/schema_v1.md

@@ -1,98 +0,0 @@
-## Database Schema (V1 Baseline)
-
-This document describes the current relational schema for the transcription system.
-
-All primary and foreign keys in the domain models are UUID-based in V1.
-
----
-
-## Schema Diagram
-
-```mermaid
-erDiagram
-    DOCUMENT {
-        UUID id PK
-        TEXT name
-    }
-
-    JOB {
-        UUID id PK
-        UUID document_id FK
-        TEXT status
-        INTEGER retry_count
-        DATETIME date_created
-        DATETIME date_updated
-        TEXT provider
-        TEXT model
-        TEXT prompt_name
-        TEXT text
-        TEXT error_detail
-    }
-
-    SOURCE {
-        UUID id PK
-        UUID document_id FK
-        UUID job_id FK
-        TEXT upload_name
-        TEXT filename
-        TEXT file_path
-        DATETIME date_uploaded
-    }
-
-    REVISION {
-        UUID id PK
-        UUID source_id "FK, UK"
-        INTEGER revision
-        TEXT text
-        DATETIME date_created
-    }
-
-    DOCUMENT ||--o{ SOURCE : has_many
-    DOCUMENT ||--o{ JOB : has_many
-    JOB ||--o{ SOURCE : referenced_by
-    SOURCE ||--o| REVISION : has_optional_one
-```
-
----
-
-## Table Relationships and Constraints
-
-- A `Document` can have zero or more `Source` records.
-- A `Document` can have zero or more `Job` records.
-- A `Source` belongs to exactly one `Document` and one `Job`.
-- A `Source` may have one optional `Revision`.
-- Optional `0..1` revision cardinality is enforced by uniqueness on `revision.source_id`.
-
-### Invariants
-
-- `Job.text` stores immutable original provider transcription output.
-- `Revision` rows are optional user-authored edits derived from original transcription.
-- Revisions do not overwrite original `Job.text`.
-- Job status lifecycle values are: `queued`, `processing`, `transcribed`, `failed`.
-
-### Timestamp Fields
-
-- `Job.date_created`
-- `Job.date_updated`
-- `Source.date_uploaded`
-- `Revision.date_created`
-
----
-
-## Related Local References
-
-- [System Overview](index_v1.md)
-- [System Design Intent](intent.md)
-- [Transcription Methodology](transcription_methodology.md)
-- [System Architecture](architecture_v1.md)
-- [System Requirements](requirements_v1.md)
-- Data model (this document)
-- [Error Handling Policy](error_handling_v1.md)
-- [Implementation Plan](implementation_plan_v1.md)
-
-## Glossary
-
-- **Document**: logical grouping for one or more transcribed sources.
-- **Source**: uploaded file content (image/PDF) linked to a job.
-- **Job**: processing record that stores lifecycle status and original output.
-- **Revision**: optional single user-authored edited text linked to a source.

docs/ver1/traceability_v1.md

@@ -1,40 +0,0 @@
-# V1 Traceability Matrix
-
-This matrix provides implementation and validation evidence for V1 requirements (`REQ-0` through `REQ-13`).
-
-Status values:
-
-- `done`: implemented and evidence recorded
-- `in progress`: partially implemented or evidence incomplete
-- `not started`: no implementation/evidence yet
-
-## Requirement Evidence Table
-
-| Requirement | Status | Implementation Evidence | Validation Evidence |
-| --- | --- | --- | --- |
-| REQ-0 | done | End-to-end upload + worker pipeline in `src/transcription/services/store.py`, `src/transcription/worker.py`, `src/transcription/services/workflows.py` | `tests/integration/test_pipeline_flow.py` |
-| REQ-1 | done | Upload UI/page flow in `src/transcription/ui/pages/upload_page.py`, `src/transcription/ui/components/upload.py` | `tests/ui/test_upload_page.py`, `tests/integration/test_pipeline_flow.py` |
-| REQ-2 | done | Async worker execution and provider call orchestration in `src/transcription/worker.py`, `src/transcription/services/workflows.py` | `tests/integration/test_pipeline_flow.py`, `tests/services/test_workflows_reliability.py` |
-| REQ-3 | done | Job lifecycle state model + transitions in `src/transcription/models.py`, `src/transcription/services/jobs.py`, `src/transcription/services/workflows.py` | `tests/services/test_job_service.py`, `tests/ui/test_jobs_page.py` |
-| REQ-4 | done | Persistence of original output and failure detail in `src/transcription/services/transcription.py`, `src/transcription/services/workflows.py` | `tests/integration/test_pipeline_flow.py`, `tests/services/test_workflows_reliability.py` |
-| REQ-5 | done | Status/result inspection via UI pages and API health route in `src/transcription/ui/pages/jobs_page.py`, `src/transcription/api/health.py` | `tests/ui/test_jobs_page.py`, `tests/ui/test_pages_registration.py`, `tests/api/test_health.py` |
-| REQ-6 | done | Background processing trigger/worker notifier and non-blocking workflow in `src/transcription/ui/components/upload.py`, `src/transcription/worker.py` | `tests/test_app.py`, `tests/services/test_workflows_reliability.py` |
-| REQ-7 | done | Lifespan-owned runtime resources in `src/transcription/app.py`, `src/transcription/db/runtime.py` | `tests/test_app.py`, `tests/test_db.py` |
-| REQ-8 | done | Centralized settings/logging initialization in `src/transcription/config.py`, `src/transcription/app.py` | `tests/test_config.py`, `tests/test_app.py` |
-| REQ-9 | done | Containerized runtime baseline in `docker-compose.yml`, `Dockerfile` | `release_checklist_v1.md` (Ops checklist), manual demonstration step |
-| REQ-10 | done | Explicit schema bootstrap policy + runtime controls in `src/transcription/config.py`, `src/transcription/app.py`, `src/transcription/db/operations.py` | `tests/test_db.py`, `tests/test_config.py` |
-| REQ-11 | done | Service/workflow persistence boundaries in `src/transcription/services/*.py`, `src/transcription/services/workflows.py` | `tests/services/test_job_service.py`, `tests/services/test_transcription_service.py` |
-| REQ-12 | done | Prompt artifacts in `prompts/` and loading/validation in `src/transcription/services/transcription.py` | `tests/test_prompts.py` |
-| REQ-13 | done | Optional single revision create/update/view/delete in `src/transcription/services/transcription.py`, `src/transcription/ui/pages/jobs_page.py` | `tests/services/test_transcription_service.py`, `tests/ui/test_jobs_page.py` |
-
-## Operational Evidence (Step 3 Artifacts)
-
-- Runbook: `runbook_v1.md`
-- Migration/backfill/rollback guidance: `migration_v1.md`
-- Release readiness checklist: `release_checklist_v1.md`
-
-## Verification Cadence
-
-- Per change: maintain `tests/test_traceability.py` mappings for touched requirements.
-- Per milestone: update this table status and evidence links.
-- Pre-release: confirm all rows are `done` and non-external suite is green.

prompts/README.md

@@ -5,9 +5,11 @@ This directory stores transcription prompts as individual Markdown artifacts.
 ## Conventions
 - Keep one prompt per file.
 - Use stable, descriptive snake_case file names.
+- Store prompt files directly in this directory; nested paths are rejected.
 - Prefer incremental edits to a single prompt per change for clean history.
 - Keep prompts human-readable and policy-focused.
 - Do not store secrets in prompt files.
+- Runtime jobs snapshot prompt text, SHA-256 provenance, and sampling configuration.
 
 ## Current Prompt
 - `transcribe_document.md`: baseline verbatim transcription policy for historical documents.

prompts/system.txt

@@ -1,10 +0,0 @@
-You are an assistant that may call tools.
-
-Tool safety rules:
-1) Tool arguments MUST be strict JSON matching the schema exactly.
-2) Never place disallowed, sensitive, explicit, or policy-violating text directly into tool arguments.
-3) If user content may be unsafe, first produce a brief neutral summary and pass only that summary.
-4) Prefer IDs, enums, booleans, and short fields over raw free-form text.
-5) Keep all string arguments <= 300 chars unless schema says otherwise.
-6) If you cannot safely provide valid tool args, do not call the tool; respond with "NO_TOOL_CALL" and explain briefly.
-7) Never include markdown/code fences in tool arguments.

prompts/transcribe_document.md

@@ -6,9 +6,15 @@ Do not summarize. Do not paraphrase. Do not modernize style.
 
 ## Output Contract
 - Return only the transcription text.
+- Begin with exactly one applicable body marker:
+  - `[document body handwritten]`
+  - `[document body typewritten]`
+  - `[document body typeset]`
+  - `[document body mixed]`
 - Preserve original wording, punctuation, and meaningful structure.
 - Keep line/section flow readable while preserving intent and document organization.
 - Never invent missing content.
+- Use ordinary plain-text characters rather than HTML entities.
 
 ## Rules for Ambiguous or Damaged Text
 
@@ -46,6 +52,31 @@ Do not summarize. Do not paraphrase. Do not modernize style.
 - Signal location before the note text.
 - Example form: `[written in left margin: ...]`
 
+### Document body medium
+- Use `[document body handwritten]` when the main body is written by hand.
+- Use `[document body typewritten]` for mechanically typewritten pages. Uneven impressions,
+  monospaced characters, worn type, and other typewriter defects are not handwriting.
+- Use `[document body typeset]` for printed pages composed with movable type or comparable
+  typesetting.
+- Use `[document body mixed]` when substantial body content uses more than one medium, such
+  as a completed printed form.
+- Preserve printed and handwritten text together in their original reading context.
+- On mixed documents, leave printed labels and instructions unmarked and wrap only actual
+  handwritten entries in `[handwritten: ...]`.
+- Mark handwritten signatures as `[handwritten signature: ...]`.
+- If the main body is entirely handwritten, use its one body marker rather than wrapping
+  each line in `[handwritten: ...]`.
+- Mark later notes or uncertain additions as `[handwritten annotation: ...]`.
+- When authorship is unclear, use `[handwritten annotation, author uncertain: ...]`.
+- Do not infer authorship, writing date, or whether different handwriting belongs to different people unless explicitly evident.
+
+### Structured layouts
+- Preserve tables of contents as associated title, dotted-leader, and page-reference rows.
+- Preserve tables and forms in reading order while keeping labels associated with their values.
+- Preserve columns in their evident reading order; do not interleave unrelated rows.
+- Preserve captions with the visual element they describe.
+- Preserve marginalia with its location marker and page numbers in their evident position.
+
 ### Line-break hyphenation
 - Rejoin words split across line breaks when they are clearly one word.
 - Remove only line-break hyphens used for wrapping.
@@ -70,3 +101,5 @@ Before finalizing, ensure:
 2. Uncertain/illegible areas are explicitly marked.
 3. Crossed-out and inserted text are preserved with required tags.
 4. Structure/ordering is preserved as faithfully as possible.
+5. Exactly one document-body marker appears, and localized handwriting markers are used only where applicable.
+6. Tables, forms, columns, captions, marginalia, dotted leaders, and page references retain their associations.

pyproject.toml

@@ -15,11 +15,17 @@ dependencies = [
     "aiosqlite>=0.21.0",
     "asyncpg>=0.31.0",
     "fastapi>=0.138.0",
+    # Exact pin, deliberate. NiceGUI 3.x minor releases ship Quasar/Vue changes that
+    # break component props and styling, and tests/ui/ cannot detect visual regressions.
+    # Hold through the current release stabilization; revisit as a scheduled upgrade.
+    # See docs/production-runbook.md, "Dependency upgrade policy".
     "nicegui==3.13.0",
     "openrouter>=0.7.0",
+    "pillow>=10.0.0",
     "psycopg2-binary>=2.9.12",
     "pydantic>=2.13.4",
     "pydantic-settings>=2.9.1",
+    "python-gedcom>=1.1.0",
     "sqlmodel>=0.0.25",
 ]
 
@@ -40,6 +46,7 @@ dev = [
 [tool.pytest.ini_options]
 addopts = "--strict-markers -q"
 asyncio_mode = "strict"
+asyncio_default_fixture_loop_scope = "function"
 filterwarnings = [
     "error:coroutine .* was never awaited:RuntimeWarning",
 ]

ruff.toml

@@ -26,6 +26,7 @@ extend-select = [
 	"E", "W",   # https://docs.astral.sh/ruff/rules/#pycodestyle-e-w
 	"F",        # https://docs.astral.sh/ruff/rules/#pyflakes-f
     "FURB",     # https://docs.astral.sh/ruff/rules/#refurb-furb
+    "G",        # https://docs.astral.sh/ruff/rules/#flake8-logging-format-g
 	"I",        # https://docs.astral.sh/ruff/rules/#isort-i
     "N",        # https://docs.astral.sh/ruff/rules/#pep8-naming-n
     "PD",       # https://docs.astral.sh/ruff/rules/#pandas-vet-pd

src/transcription/__main__.py

@@ -0,0 +1,27 @@
+import uvicorn
+from fastapi import FastAPI
+
+from .app import create_app
+from .config import parse_cli_settings
+
+
+def create_cli_app() -> FastAPI:
+    """Create an app from CLI settings for Uvicorn's reload process."""
+    return create_app(settings=parse_cli_settings())
+
+
+def main() -> None:
+    settings = parse_cli_settings()
+    application = "transcription.__main__:create_cli_app" if settings.reload else create_app(settings=settings)
+    uvicorn.run(
+        application,
+        factory=settings.reload,
+        host=settings.host,
+        port=settings.port,
+        log_level=settings.log_level,
+        reload=settings.reload,
+    )
+
+
+if __name__ == "__main__":
+    main()

src/transcription/api/documents_api.py

@@ -0,0 +1,202 @@
+"""API routes for relationship and classification registries."""
+
+from __future__ import annotations
+
+from typing import Annotated
+from uuid import UUID
+
+from fastapi import APIRouter
+from fastapi import Depends
+from fastapi import Request
+from fastapi import Response
+from pydantic import BaseModel
+from pydantic import ConfigDict
+from pydantic import Field
+
+from transcription.db.models import Document
+from transcription.db.models import DocumentPerson
+from transcription.db.models import DocumentType
+from transcription.db.models import PersonRole
+from transcription.services import DocumentService
+from transcription.services import PeopleService
+
+router = APIRouter(prefix="/api", tags=["documents"])
+
+
+class ApiModel(BaseModel):
+    model_config = ConfigDict(extra="forbid", frozen=True, str_strip_whitespace=True)
+
+
+class DocumentTypeRead(ApiModel):
+    id: UUID
+    label: str
+    is_active: bool
+
+
+class PersonRoleRead(ApiModel):
+    id: UUID
+    label: str
+    is_active: bool
+
+
+class DocumentTypeWriteRequest(ApiModel):
+    document_type_id: UUID
+
+
+class DocumentTypeWriteResponse(ApiModel):
+    document_id: UUID
+    document_type_id: UUID
+
+
+class DocumentPersonWriteRequest(ApiModel):
+    person_id: UUID
+    role_id: UUID
+
+
+class DocumentPersonRoleUpdateRequest(ApiModel):
+    role_id: UUID
+
+
+class DocumentPersonRead(ApiModel):
+    id: UUID
+    document_id: UUID
+    person_id: UUID
+    role_id: UUID
+    role_label: str | None = None
+    person_name: str | None = None
+
+
+class DocumentPeopleResponse(ApiModel):
+    document_id: UUID
+    links: list[DocumentPersonRead] = Field(default_factory=list)
+
+
+def _document_type_to_read(item: DocumentType) -> DocumentTypeRead:
+    item_id, label, is_active = _registry_read_values(item)
+    return DocumentTypeRead(id=item_id, label=label, is_active=is_active)
+
+
+def _person_role_to_read(item: PersonRole) -> PersonRoleRead:
+    item_id, label, is_active = _registry_read_values(item)
+    return PersonRoleRead(id=item_id, label=label, is_active=is_active)
+
+
+def _registry_read_values(item: DocumentType | PersonRole) -> tuple[UUID, str, bool]:
+    return item.id, item.label, item.is_active
+
+
+def _document_person_to_read(item: DocumentPerson) -> DocumentPersonRead:
+    person_name = item.person.full_name if item.person is not None else None
+    return DocumentPersonRead(
+        id=item.id,
+        document_id=item.document_id,
+        person_id=item.person_id,
+        role_id=item.role_id,
+        role_label=item.role_ref.label if item.role_ref is not None else None,
+        person_name=person_name,
+    )
+
+
+def _document_to_type_response(item: Document) -> DocumentTypeWriteResponse:
+    if item.document_type_id is None:
+        raise ValueError("Document Type assignment did not persist")
+    return DocumentTypeWriteResponse(
+        document_id=item.id,
+        document_type_id=item.document_type_id,
+    )
+
+
+def get_document_service(request: Request) -> DocumentService:
+    """Resolve the document service from app lifespan state when available."""
+    services = getattr(request.app.state, "services", None)
+    if services is not None:
+        return services.documents
+    return DocumentService()
+
+
+def get_people_service(request: Request) -> PeopleService:
+    """Resolve the People service from app lifespan state when available."""
+    services = getattr(request.app.state, "services", None)
+    if services is not None:
+        return services.people
+    return PeopleService()
+
+
+DocumentServiceDependency = Annotated[DocumentService, Depends(get_document_service)]
+PeopleServiceDependency = Annotated[PeopleService, Depends(get_people_service)]
+
+
+@router.get("/document-types", response_model=list[DocumentTypeRead])
+async def list_document_types(
+    service: DocumentServiceDependency,
+    active_only: bool = True,
+) -> list[DocumentTypeRead]:
+    items = await service.list_document_types(active_only=active_only)
+    return [_document_type_to_read(item) for item in items]
+
+
+@router.get("/person-roles", response_model=list[PersonRoleRead])
+async def list_person_roles(
+    service: PeopleServiceDependency,
+    active_only: bool = True,
+) -> list[PersonRoleRead]:
+    items = await service.list_person_roles(active_only=active_only)
+    return [_person_role_to_read(item) for item in items]
+
+
+@router.put("/documents/{document_id}/type", response_model=DocumentTypeWriteResponse)
+async def set_document_type(
+    document_id: UUID,
+    payload: DocumentTypeWriteRequest,
+    service: DocumentServiceDependency,
+) -> DocumentTypeWriteResponse:
+    document = await service.set_document_type(
+        document_id=document_id,
+        document_type_id=payload.document_type_id,
+    )
+    return _document_to_type_response(document)
+
+
+@router.get("/documents/{document_id}/people", response_model=DocumentPeopleResponse)
+async def list_document_people(
+    document_id: UUID,
+    service: PeopleServiceDependency,
+) -> DocumentPeopleResponse:
+    links = await service.list_document_people(document_id=document_id)
+    return DocumentPeopleResponse(document_id=document_id, links=[_document_person_to_read(item) for item in links])
+
+
+@router.post("/documents/{document_id}/people", response_model=DocumentPersonRead)
+async def add_document_person_link(
+    document_id: UUID,
+    payload: DocumentPersonWriteRequest,
+    service: PeopleServiceDependency,
+) -> DocumentPersonRead:
+    link = await service.add_document_person_link(
+        document_id=document_id,
+        person_id=payload.person_id,
+        role_id=payload.role_id,
+    )
+    return _document_person_to_read(link)
+
+
+@router.patch("/document-people/{document_person_id}", response_model=DocumentPersonRead)
+async def set_document_person_role(
+    document_person_id: UUID,
+    payload: DocumentPersonRoleUpdateRequest,
+    service: PeopleServiceDependency,
+) -> DocumentPersonRead:
+    link = await service.set_document_person_role(
+        document_person_id=document_person_id,
+        role_id=payload.role_id,
+    )
+    return _document_person_to_read(link)
+
+
+@router.delete("/document-people/{document_person_id}", status_code=204)
+async def delete_document_person_link(
+    document_person_id: UUID,
+    service: PeopleServiceDependency,
+) -> Response:
+    await service.remove_document_person_link(document_person_id=document_person_id)
+    return Response(status_code=204)

src/transcription/api/errors.py

@@ -21,7 +21,9 @@ _STATUS_BY_CATEGORY: dict[ErrorCategory, int] = {
     ErrorCategory.NOT_FOUND: 404,
     ErrorCategory.CONFLICT: 409,
     ErrorCategory.EXTERNAL_PROVIDER: 503,
+    ErrorCategory.EXTERNAL_TIMEOUT: 503,
     ErrorCategory.INFRA_TRANSIENT: 503,
+    ErrorCategory.PROCESSING: 500,
     ErrorCategory.INFRA_PERSISTENT: 500,
     ErrorCategory.INTERNAL_UNEXPECTED: 500,
 }

src/transcription/api/health.py

@@ -1,16 +1,44 @@
 """Health endpoint routes."""
 
+from typing import NotRequired
+from typing import TypedDict
+
 from fastapi import APIRouter
+from fastapi import Request
+
+from transcription.worker import resolve_worker_health
 
 router = APIRouter()
 
 
-def healthz() -> dict[str, str]:
-    """Return a simple health status payload."""
-    return {"status": "ok"}
+class WorkerHealthPayload(TypedDict):
+    state: str
+    error_id: NotRequired[str]
+    error_category: NotRequired[str]
+
+
+class HealthPayload(TypedDict):
+    status: str
+    worker: WorkerHealthPayload
+
+
+def healthz(request: Request) -> HealthPayload:
+    """Return health status with worker-liveness signal."""
+    worker = resolve_worker_health(request.app.state)
+    payload: HealthPayload = {
+        "status": "ok",
+        "worker": {
+            "state": worker.state,
+        },
+    }
+    if worker.error_id is not None:
+        payload["worker"]["error_id"] = worker.error_id
+    if worker.error_category is not None:
+        payload["worker"]["error_category"] = worker.error_category
+    return payload
 
 
 @router.get("/healthz")
-def healthz_route() -> dict[str, str]:
+def healthz_route(request: Request) -> HealthPayload:
     """Route wrapper for health status payload."""
-    return healthz()
+    return healthz(request)

src/transcription/api/print_api.py

@@ -0,0 +1,54 @@
+"""Safe media route for Document print previews."""
+
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Annotated
+from uuid import UUID
+
+from fastapi import APIRouter
+from fastapi import Depends
+from fastapi import HTTPException
+from fastapi import Request
+from fastapi.responses import FileResponse
+
+from transcription.services.source_media import SOURCE_MIME_TYPES
+from transcription.services.sources import SourceService
+
+router = APIRouter(prefix="/api", tags=["print"])
+
+
+def get_source_service(request: Request) -> SourceService:
+    services = getattr(request.app.state, "services", None)
+    if services is not None:
+        return services.sources
+    return SourceService()
+
+
+SourceServiceDependency = Annotated[SourceService, Depends(get_source_service)]
+
+
+@router.get("/documents/{document_id}/sources/{source_id}/media", response_class=FileResponse)
+async def read_document_source_media(
+    document_id: UUID,
+    source_id: UUID,
+    service: SourceServiceDependency,
+) -> FileResponse:
+    """Serve one validated Source through record identifiers, never a supplied path."""
+    source = await service.read_source(source_id)
+    if source.document_id != document_id:
+        raise HTTPException(status_code=404, detail="Source not found for Document")
+
+    upload_root = service.settings.upload_dir.resolve()
+    path = (upload_root / Path(source.file_path)).resolve()
+    try:
+        path.relative_to(upload_root)
+    except ValueError as exc:
+        raise HTTPException(status_code=404, detail="Source media is outside managed storage") from exc
+    if not path.is_file():
+        raise HTTPException(status_code=404, detail="Source media is unavailable")
+
+    media_type = SOURCE_MIME_TYPES.get(path.suffix.lower())
+    if media_type is None:
+        raise HTTPException(status_code=415, detail="Unsupported Source media type")
+    return FileResponse(path, media_type=media_type)

src/transcription/app.py

@@ -14,15 +14,19 @@ from fastapi import status
 from fastapi.responses import RedirectResponse
 from fastapi.staticfiles import StaticFiles
 
+from .api.documents_api import router as documents_router
 from .api.errors import register_error_handlers
 from .api.health import router as health_router
+from .api.print_api import router as print_router
+from .config import Settings
 from .config import configure_logging
 from .config import get_settings
 from .db import create_all
 from .db import dispose_database_runtime
 from .db import initialize_database_runtime
+from .db import reconcile_canonical_media_paths
+from .db import reconcile_legacy_job_source_columns
 from .services import ServiceBundle
-from .services.jobs import JobService
 from .ui import register_pages
 from .worker import worker_consumer_lifespan
 
@@ -31,68 +35,77 @@ logger = logging.getLogger(__name__)
 
 @asynccontextmanager
 async def _lifespan(app: FastAPI):
-    configure_logging()
-
     settings = getattr(app.state, "settings", None) or get_settings()
+    configure_logging(settings)
     app.state.settings = settings
-    app.state.services = ServiceBundle()
     app.state.runtime = initialize_database_runtime(settings=settings)
+    session_factory = app.state.runtime.session_factory
+    app.state.services = ServiceBundle.from_session_factory(session_factory, settings=settings)
 
     if settings.should_bootstrap_schema:
         await create_all(engine=app.state.runtime.engine)
+    await reconcile_legacy_job_source_columns(engine=app.state.runtime.engine)
+    await reconcile_canonical_media_paths(engine=app.state.runtime.engine)
 
     settings.upload_dir.mkdir(parents=True, exist_ok=True)
     settings.prompt_dir.mkdir(parents=True, exist_ok=True)
+    settings.log_dir.mkdir(parents=True, exist_ok=True)
 
     await _recover_stale_processing_jobs(app)
 
     async with AsyncExitStack() as stack:
         stack.push_async_callback(dispose_database_runtime)
-        stop_event, worker_notifier = await stack.enter_async_context(
+        if settings.run_embedded_worker:
+            stop_event, worker_notifier, worker_health = await stack.enter_async_context(
                 worker_consumer_lifespan(
                     session_factory=app.state.runtime.session_factory,
-                poll_interval_seconds=1.0,
+                    poll_interval_seconds=settings.worker_poll_interval_seconds,
+                    shutdown_timeout_seconds=(
+                        settings.worker_provider_timeout_seconds + settings.worker_shutdown_grace_seconds
+                    ),
                 )
             )
             app.state.worker_stop_event = stop_event
             app.state.worker_notifier = worker_notifier
+            app.state.worker_health = worker_health
         yield
 
 
 async def _recover_stale_processing_jobs(app: FastAPI) -> None:
     """Re-queue stale processing jobs at startup.
 
-    Any job left in PROCESSING longer than the configured provider timeout is
-    assumed orphaned and moved back to QUEUED before the worker starts.
+    Any job left in PROCESSING longer than the stale-job threshold is assumed
+    orphaned and moved back to QUEUED before the worker starts.
     """
     settings = app.state.settings
-    stale_before = datetime.now(UTC) - timedelta(seconds=settings.worker_provider_timeout_seconds)
-    job_service = JobService(session_factory=app.state.runtime.session_factory)
-    recovered = await job_service.requeue_stale_processing_jobs(stale_before=stale_before)
+    stale_before = datetime.now(UTC) - timedelta(seconds=settings.worker_stale_job_seconds)
+    recovered = await app.state.services.jobs.requeue_stale_processing_jobs(stale_before=stale_before)
     if recovered > 0:
         logger.warning("Recovered %s stale processing job(s) at startup", recovered)
 
 
-def create_app() -> FastAPI:
+def create_app(settings: Settings | None = None) -> FastAPI:
     """Create and configure the FastAPI application."""
     app = FastAPI(title="Transcription", lifespan=_lifespan)
-    settings = get_settings()
-    app.state.settings = settings
+    active_settings = settings or get_settings()
+    app.state.settings = active_settings
     app.mount(
         "/uploads",
-        StaticFiles(directory=settings.upload_dir, check_dir=False),
+        StaticFiles(directory=active_settings.upload_dir, check_dir=False),
         name="uploads",
     )
 
     @app.get("/", include_in_schema=False)
     async def root_redirect() -> RedirectResponse:
-        return RedirectResponse(url="/ui", status_code=status.HTTP_307_TEMPORARY_REDIRECT)
+        return RedirectResponse(url="/ui/homepage", status_code=status.HTTP_307_TEMPORARY_REDIRECT)
 
     @app.get("/ui", include_in_schema=False)
     async def ui_redirect() -> RedirectResponse:
-        return RedirectResponse(url="/ui/upload", status_code=status.HTTP_307_TEMPORARY_REDIRECT)
+        return RedirectResponse(url="/ui/homepage", status_code=status.HTTP_307_TEMPORARY_REDIRECT)
 
     register_error_handlers(app)
-    register_pages(app)
     app.include_router(health_router)
+    app.include_router(documents_router)
+    app.include_router(print_router)
+    register_pages(app)
     return app

src/transcription/app_state.py

@@ -1,39 +0,0 @@
-"""Helpers for accessing lifespan-owned application state resources."""
-
-from __future__ import annotations
-
-from fastapi import FastAPI
-from sqlalchemy.ext.asyncio import async_sessionmaker
-from sqlmodel.ext.asyncio.session import AsyncSession
-
-from transcription.db.runtime import DatabaseRuntime
-from transcription.db.runtime import get_session_factory
-from transcription.worker import WorkerNotifier
-from transcription.worker import resolve_worker_notifier
-
-
-def resolve_database_runtime(state: object) -> DatabaseRuntime | None:
-    """Return database runtime from app-like state objects when available."""
-    runtime = getattr(state, "runtime", None)
-    return runtime if isinstance(runtime, DatabaseRuntime) else None
-
-
-def require_database_runtime(state: object) -> DatabaseRuntime:
-    """Return database runtime or raise when app lifespan has not initialized it."""
-    runtime = resolve_database_runtime(state)
-    if runtime is None:
-        raise RuntimeError("Database runtime is not initialized on application state")
-    return runtime
-
-
-def resolve_session_factory(state: object) -> async_sessionmaker[AsyncSession]:
-    """Return DB session factory from state when available, otherwise shared runtime."""
-    runtime = resolve_database_runtime(state)
-    if runtime is not None:
-        return runtime.session_factory
-    return get_session_factory()
-
-
-def get_worker_notifier(app: FastAPI) -> WorkerNotifier:
-    """Return app worker notifier, or a no-op fallback when unavailable."""
-    return resolve_worker_notifier(app.state)

src/transcription/benchmarking.py

@@ -0,0 +1,91 @@
+"""Private-corpus benchmark contracts and deterministic text scoring.
+
+This module is retained as the implementation of the evaluation policy in
+`docs/invariant/ai_evidence_and_provenance.md` §5. Application runtime paths do
+not call it directly, but preserved execution attempts and manually reviewed
+references need a deterministic scorer that remains importable for tests and
+operator tooling.
+"""
+
+from __future__ import annotations
+
+from uuid import UUID
+
+from pydantic import BaseModel
+from pydantic import ConfigDict
+from pydantic import Field
+
+
+class BenchmarkModel(BaseModel):
+    model_config = ConfigDict(extra="forbid", frozen=True)
+
+
+class EditorialAssessment(BenchmarkModel):
+    """Manually reviewed errors not represented adequately by CER or WER."""
+
+    omissions: int = Field(default=0, ge=0)
+    inventions: int = Field(default=0, ge=0)
+    silent_normalizations: int = Field(default=0, ge=0)
+    uncertainty_errors: int = Field(default=0, ge=0)
+    layout_errors: int = Field(default=0, ge=0)
+
+
+class BenchmarkScore(BenchmarkModel):
+    """Measured score for one preserved execution attempt."""
+
+    execution_attempt_id: UUID
+    character_error_rate: float = Field(ge=0)
+    word_error_rate: float = Field(ge=0)
+    character_edits: int = Field(ge=0)
+    word_edits: int = Field(ge=0)
+    reference_characters: int = Field(ge=0)
+    reference_words: int = Field(ge=0)
+    assessment: EditorialAssessment
+    latency_ms: int = Field(ge=0)
+    cost_usd: float | None = Field(default=None, ge=0)
+
+
+def score_transcription(
+    *,
+    execution_attempt_id: UUID,
+    reference: str,
+    candidate: str,
+    assessment: EditorialAssessment,
+    latency_ms: int,
+    cost_usd: float | None = None,
+) -> BenchmarkScore:
+    """Score literal text without case-folding or silent normalization."""
+    reference_words = reference.split()
+    candidate_words = candidate.split()
+    character_edits = _levenshtein(list(reference), list(candidate))
+    word_edits = _levenshtein(reference_words, candidate_words)
+    return BenchmarkScore(
+        execution_attempt_id=execution_attempt_id,
+        character_error_rate=character_edits / max(1, len(reference)),
+        word_error_rate=word_edits / max(1, len(reference_words)),
+        character_edits=character_edits,
+        word_edits=word_edits,
+        reference_characters=len(reference),
+        reference_words=len(reference_words),
+        assessment=assessment,
+        latency_ms=latency_ms,
+        cost_usd=cost_usd,
+    )
+
+
+def _levenshtein(reference: list[str], candidate: list[str]) -> int:
+    if len(reference) < len(candidate):
+        reference, candidate = candidate, reference
+    previous = list(range(len(candidate) + 1))
+    for reference_index, reference_value in enumerate(reference, start=1):
+        current = [reference_index]
+        for candidate_index, candidate_value in enumerate(candidate, start=1):
+            current.append(
+                min(
+                    current[-1] + 1,
+                    previous[candidate_index] + 1,
+                    previous[candidate_index - 1] + (reference_value != candidate_value),
+                )
+            )
+        previous = current
+    return previous[-1]

src/transcription/config.py

@@ -1,81 +1,240 @@
 """Centralized application configuration.
 
-All settings are loaded from environment variables (or a .env file)
+All settings are loaded from environment variables (or an env file)
 once at startup. Provider-specific defaults (model names, base URLs)
 are resolved by the provider adapters, not here.
 """
 
+import copy
 import logging.config
-from contextvars import ContextVar
+import os
+from collections.abc import Sequence
 from enum import StrEnum
+from functools import cache
 from pathlib import Path
+from typing import Annotated
+from typing import Any
 from typing import Literal
 
+from pydantic import BaseModel
+from pydantic import ConfigDict
 from pydantic import Field
+from pydantic import SecretStr
+from pydantic import StringConstraints
+from pydantic import field_validator
+from pydantic import model_validator
 from pydantic_settings import BaseSettings
 from pydantic_settings import SettingsConfigDict
 
 logger = logging.getLogger(__name__)
+PROJECT_ROOT = Path(__file__).resolve().parents[2]
+DEFAULT_ENV_FILE_NAME = ".env.production"
+
+
+def resolve_settings_env_file_path() -> Path:
+    """Resolve the runtime env file independent of the process working directory."""
+    override = os.getenv("ENV_FILE", "").strip()
+    if override:
+        return Path(override)
+    return PROJECT_ROOT / DEFAULT_ENV_FILE_NAME
 
 
 class Provider(StrEnum):
     OPENROUTER = "openrouter"
 
 
+NonEmptyStr = Annotated[str, StringConstraints(strip_whitespace=True, min_length=1)]
+PromptFilename = Annotated[str, StringConstraints(strip_whitespace=True, min_length=1, pattern=r"^[^/\\]+$")]
+Probability = Annotated[float, Field(ge=0.0, le=1.0)]
+Temperature = Annotated[float, Field(ge=0.0, le=2.0)]
+DEFAULT_PROVIDER_MODEL = "google/gemini-2.5-flash"
+WORKER_STALE_TIMEOUT_MULTIPLIER = 3.0
+
+
+class SqliteSettings(BaseModel):
+    model_config = ConfigDict(extra="forbid", frozen=True)
+
+    driver: Literal["sqlite"] = "sqlite"
+    path: NonEmptyStr = "./data/transcription.db"
+
+
+class PostgresSettings(BaseModel):
+    model_config = ConfigDict(extra="forbid", frozen=True)
+
+    driver: Literal["postgres"] = "postgres"
+    host: NonEmptyStr
+    port: int = Field(default=5432, ge=1, le=65535)
+    database: NonEmptyStr
+    user: NonEmptyStr
+    password: SecretStr
+
+
+DatabaseSettings = Annotated[
+    SqliteSettings | PostgresSettings,
+    Field(discriminator="driver"),
+]
+
+
 class Settings(BaseSettings):
     model_config = SettingsConfigDict(
-        env_file=".env",
+        env_file=None,
         env_file_encoding="utf-8",
         extra="ignore",
+        env_nested_delimiter="__",
+        cli_implicit_flags=True,
+        cli_kebab_case=True,
+        frozen=True,
     )
 
+    def __init__(self, /, **values: Any) -> None:
+        if "_env_file" not in values:
+            values["_env_file"] = resolve_settings_env_file_path()
+        super().__init__(**values)
+
+    # --- NiceGUI Server ---
+    host: str = "0.0.0.0"
+    port: int = 8000
+    log_level: Literal["critical", "error", "warning", "info", "debug", "trace"] = "info"
+    reload: bool = False
+    log_dir: Path = Path("./data/logs")
+    log_file_name: NonEmptyStr = "transcription.log"
+    log_file_max_bytes: int = Field(default=10 * 1024 * 1024, gt=0)
+    log_file_backup_count: int = Field(default=5, ge=1)
+
     # --- AI provider ---
     provider: Provider = Provider.OPENROUTER
-    openrouter_api_key: str
-    provider_model: str | None = None
-    openrouter_http_referer: str | None = None
-    openrouter_app_title: str | None = None
+    openrouter_api_key: SecretStr
+    provider_model: NonEmptyStr | None = DEFAULT_PROVIDER_MODEL
+    provider_models: tuple[NonEmptyStr, ...] = ()
+    openrouter_http_referer: NonEmptyStr | None = None
+    openrouter_app_title: NonEmptyStr | None = None
+    default_prompt_name: PromptFilename = "transcribe_document.md"
+    transcription_temperature: Temperature | None = None
+    transcription_top_p: Probability | None = None
 
     # --- runtime environment ---
     environment: Literal["development", "test", "production"] = "development"
+    transcription_commit: NonEmptyStr | None = None
+    run_embedded_worker: bool = True
 
     # --- persistence ---
-    database_url: str = "sqlite:///./transcription.db"
-    bootstrap_schema_on_startup: bool | None = None
+    database: DatabaseSettings = Field(default_factory=SqliteSettings)
+    bootstrap_schema_on_startup: bool = False
     sqlite_check_same_thread: bool = False
 
     # --- filesystem paths ---
-    upload_dir: Path = Path("./uploads")
+    upload_dir: Path = Path("./data")
     prompt_dir: Path = Path("./prompts")
 
     # --- worker reliability ---
-    worker_max_retries: int = 0
-    worker_retry_backoff_seconds: float = 0.0
-    worker_provider_timeout_seconds: float = Field(default=20.0, gt=0.0, le=20.0)
+    worker_max_retries: int = Field(default=0, ge=0)
+    # Bounded only from below. Vision transcription of a dense page routinely runs
+    # well past twenty seconds, so an upper cap here would silently fail real work.
+    worker_provider_timeout_seconds: float = Field(default=30.0, gt=0.0)
+    worker_stale_job_seconds: float = Field(default=90.0, gt=0.0)
+    worker_retry_backoff_seconds: float = Field(default=1.0, ge=0.0)
+    worker_shutdown_grace_seconds: float = Field(default=5.0, ge=0.0)
+    worker_poll_interval_seconds: float = Field(default=1.0, gt=0.0)
     worker_min_transcription_chars: int = Field(default=0, ge=0)
     worker_min_transcription_lines: int = Field(default=0, ge=0)
     worker_fail_on_finish_reason_length: bool = False
 
+    @field_validator("provider_models", mode="before")
+    @classmethod
+    def validate_provider_models_input(cls, value: object) -> object:
+        if value is None:
+            return ()
+        if isinstance(value, (list, tuple)) and not value:
+            raise ValueError("PROVIDER_MODELS must contain at least one model")
+        return value
+
+    @model_validator(mode="before")
+    @classmethod
+    def normalize_provider_models(cls, data: object) -> object:
+        """Build the immutable model selector with the configured default first.
+
+        This runs before field validation so the derived value is produced by
+        normal construction rather than by mutating a frozen instance.
+        """
+        if not isinstance(data, dict):
+            return data
+
+        default_model = data.get("provider_model") or DEFAULT_PROVIDER_MODEL
+        if not isinstance(default_model, str):
+            return data
+        default_model = default_model.strip()
+
+        configured = data.get("provider_models")
+        if configured is None:
+            configured = ()
+        elif isinstance(configured, str):
+            # Left as-is so the field validator can report the malformed value.
+            return {**data, "provider_model": default_model}
+        elif not isinstance(configured, (list, tuple)):
+            return {**data, "provider_model": default_model}
+        elif not configured:
+            # Preserved so validate_provider_models_input can reject it.
+            return {**data, "provider_model": default_model}
+
+        deduplicated: list[str] = []
+        for model in (default_model, *configured):
+            if not isinstance(model, str):
+                return {**data, "provider_model": default_model}
+            normalized = model.strip()
+            if normalized not in deduplicated:
+                deduplicated.append(normalized)
+
+        return {**data, "provider_model": default_model, "provider_models": tuple(deduplicated)}
+
+    @model_validator(mode="before")
+    @classmethod
+    def _derive_worker_stale_job_seconds(cls, data: object) -> object:
+        """Default stale-job recovery with margin over one provider timeout."""
+        if not isinstance(data, dict):
+            return data
+        if data.get("worker_stale_job_seconds") is not None:
+            return data
+
+        timeout = data.get("worker_provider_timeout_seconds", 30.0)
+        if not isinstance(timeout, (str, int, float)):
+            return data
+        try:
+            timeout_seconds = float(timeout)
+        except ValueError:
+            return data
+        return {
+            **data,
+            "worker_stale_job_seconds": timeout_seconds * WORKER_STALE_TIMEOUT_MULTIPLIER,
+        }
+
+    @model_validator(mode="after")
+    def _validate_worker_stale_job_seconds(self) -> "Settings":
+        """Reject stale recovery that can fire before one provider timeout expires."""
+        if self.worker_stale_job_seconds <= self.worker_provider_timeout_seconds:
+            raise ValueError("WORKER_STALE_JOB_SECONDS must exceed WORKER_PROVIDER_TIMEOUT_SECONDS")
+        return self
+
     @property
     def should_bootstrap_schema(self) -> bool:
         """Return whether startup should auto-create schema for this environment."""
-        if self.bootstrap_schema_on_startup is not None:
+        if "bootstrap_schema_on_startup" in self.model_fields_set:
             return self.bootstrap_schema_on_startup
         return self.environment in {"development", "test"}
 
 
-_settings: ContextVar[Settings | None] = ContextVar("settings", default=None)
+@cache
+def get_settings(**kwargs: Any) -> Settings:
+    """Load cached settings without reading process CLI arguments."""
+    return Settings(_cli_parse_args=False, **kwargs)
 
 
-def get_settings(**kwargs) -> Settings:
-    settings = _settings.get()
-    if settings is None:
-        settings = Settings(**kwargs)  # pyright: ignore[reportCallIssue]
-        _settings.set(settings)
-    return settings
+def parse_cli_settings(args: Sequence[str] | None = None) -> Settings:
+    """Load settings with CLI arguments at the executable boundary."""
+    cli_args = True if args is None else list(args)
+    return Settings(_cli_parse_args=cli_args)
 
 
-LOGGING_CONFIG: dict[str, object] = {
+LOGGING_CONFIG: dict[str, Any] = {
     "version": 1,
     "disable_existing_loggers": False,
     "formatters": {
@@ -89,23 +248,39 @@ LOGGING_CONFIG: dict[str, object] = {
             "class": "logging.StreamHandler",
             "formatter": "standard",
             "stream": "ext://sys.stdout",
-        }
+        },
+        "file": {
+            "class": "logging.handlers.RotatingFileHandler",
+            "formatter": "standard",
+            "filename": str(Path("./data/logs") / "transcription.log"),
+            "maxBytes": 10 * 1024 * 1024,
+            "backupCount": 5,
+            "encoding": "utf-8",
+        },
     },
     "root": {
         "level": "INFO",
-        "handlers": ["console"],
+        "handlers": ["console", "file"],
     },
     "loggers": {
         "transcription": {
             "level": "DEBUG",
-            "handlers": ["console"],
+            "handlers": ["console", "file"],
             "propagate": False,
         }
     },
 }
 
 
-def configure_logging() -> None:
+def configure_logging(settings: Settings | None = None) -> None:
     """Configure root logging once at startup."""
-    logging.config.dictConfig(LOGGING_CONFIG)
+    cfg = copy.deepcopy(LOGGING_CONFIG)
+    active_settings = settings or get_settings()
+    active_settings.log_dir.mkdir(parents=True, exist_ok=True)
+    file_handler = cfg["handlers"]["file"]
+    file_handler["filename"] = str(active_settings.log_dir / active_settings.log_file_name)
+    file_handler["maxBytes"] = active_settings.log_file_max_bytes
+    file_handler["backupCount"] = active_settings.log_file_backup_count
+    cfg["loggers"]["transcription"]["level"] = active_settings.log_level.upper()
+    logging.config.dictConfig(cfg)
     logger.debug("Logging configured")

src/transcription/db/__init__.py

@@ -1,6 +1,19 @@
 from .operations import create_all
+from .operations import reconcile_canonical_media_paths
+from .operations import reconcile_legacy_job_source_columns
+from .operations import reconcile_person_name_columns
 from .runtime import dispose_database_runtime
-from .runtime import get_session
 from .runtime import initialize_database_runtime
+from .session import session_scope
+from .session import transaction_scope
 
-__all__ = ["create_all", "dispose_database_runtime", "get_session", "initialize_database_runtime"]
+__all__ = [
+    "create_all",
+    "dispose_database_runtime",
+    "initialize_database_runtime",
+    "reconcile_canonical_media_paths",
+    "reconcile_legacy_job_source_columns",
+    "reconcile_person_name_columns",
+    "session_scope",
+    "transaction_scope",
+]

src/transcription/db/engine.py

@@ -0,0 +1,75 @@
+from typing import Any
+
+from sqlalchemy import URL
+from sqlalchemy import StaticPool
+from sqlalchemy.ext.asyncio import AsyncEngine
+from sqlalchemy.ext.asyncio import create_async_engine
+
+from ..config import PostgresSettings
+from ..config import Settings
+from ..config import SqliteSettings
+from ..config import get_settings
+
+
+def get_database_url(settings: Settings) -> str:
+    match settings.database:
+        case SqliteSettings(path=path):
+            url = URL.create(
+                drivername="sqlite+aiosqlite",
+                database=path,
+            )
+        case PostgresSettings() as database:
+            url = URL.create(
+                drivername="postgresql+asyncpg",
+                host=database.host,
+                port=database.port,
+                database=database.database,
+                username=database.user,
+                password=database.password.get_secret_value(),
+            )
+    return url.render_as_string(hide_password=False)
+
+
+def resolve_engine(settings: Settings | None = None) -> AsyncEngine:
+    active_settings = settings or get_settings()
+    return get_engine(
+        get_database_url(active_settings),
+        sqlite_check_same_thread=active_settings.sqlite_check_same_thread,
+    )
+
+
+_ENGINES: dict[str, AsyncEngine] = {}
+
+
+def _create_engine(database_url: str, *, sqlite_check_same_thread: bool) -> AsyncEngine:
+    kwargs: dict[str, Any] = {"echo": False, "pool_pre_ping": True}
+    if database_url.startswith("sqlite"):
+        kwargs["connect_args"] = {"check_same_thread": sqlite_check_same_thread}
+        if ":memory:" in database_url:
+            kwargs["poolclass"] = StaticPool
+
+    return create_async_engine(database_url, **kwargs)
+
+
+def get_engine(database_url: str, *, sqlite_check_same_thread: bool = False) -> AsyncEngine:
+    """Return the process-wide engine for ``database_url``, creating it on first use.
+
+    Engines are registered per URL so that disposing one leaves every other
+    database untouched.
+    """
+    engine = _ENGINES.get(database_url)
+    if engine is None:
+        engine = _create_engine(database_url, sqlite_check_same_thread=sqlite_check_same_thread)
+        _ENGINES[database_url] = engine
+    return engine
+
+
+async def dispose_engine(database_url: str) -> None:
+    """Dispose and unregister the engine for ``database_url`` only.
+
+    Unknown URLs are a no-op rather than provoking the creation of an engine
+    purely so that it can be thrown away.
+    """
+    engine = _ENGINES.pop(database_url, None)
+    if engine is not None:
+        await engine.dispose()

src/transcription/db/loading.py

@@ -0,0 +1,45 @@
+"""Typed loader-option wrappers for SQLModel relationship attributes.
+
+SQLModel declares relationships with their runtime Python type, so
+``Document.jobs`` is annotated ``list[Job]`` even though at runtime it is an
+``InstrumentedAttribute``. SQLAlchemy's loader options are typed against
+``QueryableAttribute``, so every eager-load call site reads as a type error to a
+static checker even though the code is correct.
+
+These wrappers put that reinterpretation in one documented place instead of
+scattering a suppression comment across every eager-load call. Import
+``selectinload`` and ``defer`` from here rather than from ``sqlalchemy.orm``.
+
+Multi-level eager loads must keep using the chained form --
+``selectinload(A.b).selectinload(orm_attribute(B.c))`` -- and not the varargs
+form ``selectinload(A.b, B.c)``. The two produce the same loader path, but
+varargs applies the selectin strategy only to the last element while the
+intermediate falls back to its default strategy. Every relationship here
+declares ``lazy="raise"``, so the varargs form raises at render time.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+from typing import cast
+
+from sqlalchemy.orm import defer as _defer
+from sqlalchemy.orm import selectinload as _selectinload
+from sqlalchemy.orm.attributes import QueryableAttribute
+from sqlalchemy.orm.strategy_options import _AbstractLoad
+
+
+def orm_attribute(attribute: object) -> QueryableAttribute[Any]:
+    """Reinterpret a SQLModel relationship or field as its ORM descriptor."""
+    return cast("QueryableAttribute[Any]", attribute)
+
+
+def selectinload(*keys: object) -> _AbstractLoad:
+    """``sqlalchemy.orm.selectinload`` accepting SQLModel-annotated attributes."""
+    return _selectinload(*(orm_attribute(key) for key in keys))
+
+
+def defer(*keys: object, raiseload: bool = False) -> _AbstractLoad:
+    """``sqlalchemy.orm.defer`` accepting SQLModel-annotated attributes."""
+    first, *rest = (orm_attribute(key) for key in keys)
+    return _defer(first, *rest, raiseload=raiseload)

src/transcription/db/migration.py

@@ -0,0 +1,610 @@
+from __future__ import annotations
+
+import base64
+import json
+import shutil
+from collections.abc import Sequence
+from dataclasses import dataclass
+from datetime import UTC
+from datetime import date
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+from uuid import UUID
+from uuid import uuid4
+
+from sqlalchemy import URL
+from sqlalchemy import MetaData
+from sqlalchemy import Table
+from sqlalchemy import bindparam
+from sqlalchemy import create_engine
+from sqlalchemy import func
+from sqlalchemy import inspect as sqlalchemy_inspect
+from sqlalchemy import select
+from sqlalchemy import text
+from sqlalchemy.engine import RowMapping
+from sqlalchemy.engine import make_url
+from sqlmodel import SQLModel
+
+from transcription.config import Settings
+from transcription.config import get_settings
+
+# Register table metadata.
+from transcription.db import models as _models  # noqa: F401
+from transcription.db.engine import get_database_url
+
+EXPORT_TABLE_ORDER = (
+    "document_type",
+    "person_role",
+    "tag",
+    "document",
+    "person",
+    "genealogy_person",
+    "genealogy_family",
+    "photo",
+    "document_person",
+    "document_tag",
+    "person_tag",
+    "job",
+    "source",
+    "job_source",
+    "execution_attempt",
+    "genealogy_family_child",
+    "genealogy_citation",
+)
+
+BYTES_FIELDS = {"transport_body"}
+VERIFICATION_TABLES = EXPORT_TABLE_ORDER
+
+
+@dataclass(frozen=True)
+class MigrationPaths:
+    source_db_url: str
+    target_db_url: str
+    source_upload_dir: Path
+    target_upload_dir: Path
+    bundle_dir: Path
+
+
+def export_bundle(*, source_db_url: str, source_upload_dir: Path, bundle_dir: Path) -> None:
+    bundle_dir.mkdir(parents=True, exist_ok=True)
+    export_json = bundle_dir / "database.json"
+    uploads_bundle_dir = bundle_dir / "uploads"
+
+    payload: dict[str, Any] = {
+        "schema_name": "transcription.export-import",
+        "schema_version": "1",
+        "created_at": datetime.now(UTC).isoformat(),
+        "tables": {},
+    }
+
+    engine = create_engine(source_db_url)
+    legacy_portrait_rows: Sequence[RowMapping] = ()
+    try:  # noqa: PLR1702
+        inspector = sqlalchemy_inspect(engine)
+        source_tables = set(inspector.get_table_names())
+        metadata = MetaData()
+        metadata.reflect(bind=engine)
+        current_metadata = SQLModel.metadata
+
+        with engine.connect() as connection:
+            for table_name in EXPORT_TABLE_ORDER:
+                if table_name not in source_tables:
+                    payload["tables"][table_name] = []
+                    continue
+
+                source_table = metadata.tables[table_name]
+                target_table = current_metadata.tables[table_name]
+                export_columns = [column.name for column in target_table.columns if column.name in source_table.columns]
+                if table_name == "person" and "full_name" in source_table.columns:
+                    for legacy_column in ("full_name",):
+                        if legacy_column not in export_columns:
+                            export_columns.append(legacy_column)
+                if table_name == "person" and "portrait_path" in source_table.columns:
+                    legacy_portrait_rows = (
+                        connection.execute(
+                            select(source_table.c["id"], source_table.c["portrait_path"]).where(
+                                source_table.c["portrait_path"].is_not(None)
+                            )
+                        )
+                        .mappings()
+                        .all()
+                    )
+                rows = connection.execute(select(*(source_table.c[name] for name in export_columns))).mappings().all()
+                payload["tables"][table_name] = [
+                    _serialize_row(row, table_name=table_name, source_upload_dir=source_upload_dir) for row in rows
+                ]
+    finally:
+        engine.dispose()
+
+    if uploads_bundle_dir.exists():
+        shutil.rmtree(uploads_bundle_dir)
+    if source_upload_dir.exists():
+        shutil.copytree(source_upload_dir, uploads_bundle_dir)
+    else:
+        uploads_bundle_dir.mkdir(parents=True, exist_ok=True)
+
+    _prepare_photo_payload_and_uploads(
+        payload=payload,
+        uploads_bundle_dir=uploads_bundle_dir,
+        legacy_portrait_rows=legacy_portrait_rows,
+    )
+    _relocate_homepage_markdown(uploads_bundle_dir=uploads_bundle_dir)
+
+    export_json.write_text(json.dumps(payload, indent=2), encoding="utf-8")
+
+
+def import_bundle(*, target_db_url: str, target_upload_dir: Path, bundle_dir: Path) -> None:
+    export_json = bundle_dir / "database.json"
+    uploads_bundle_dir = bundle_dir / "uploads"
+    payload = json.loads(export_json.read_text(encoding="utf-8"))
+
+    if target_upload_dir.exists():
+        shutil.rmtree(target_upload_dir)
+    target_upload_dir.mkdir(parents=True, exist_ok=True)
+    if uploads_bundle_dir.exists():
+        shutil.copytree(uploads_bundle_dir, target_upload_dir, dirs_exist_ok=True)
+
+    _reset_sqlite_target_file(target_db_url)
+    _ensure_sqlite_target_parent_exists(target_db_url)
+    engine = create_engine(target_db_url)
+    try:
+        SQLModel.metadata.create_all(engine)
+        execution_attempt_ids = _collect_execution_attempt_ids(payload)
+        with engine.begin() as connection:
+            for table_name in reversed(EXPORT_TABLE_ORDER):
+                table = SQLModel.metadata.tables[table_name]
+                connection.execute(table.delete())
+
+            deferred_source_preferred_attempt_updates: list[dict[str, Any]] = []
+            for table_name in EXPORT_TABLE_ORDER:
+                rows = payload.get("tables", {}).get(table_name, [])
+                if not rows:
+                    continue
+                table = SQLModel.metadata.tables[table_name]
+                if table_name == "source":
+                    prepared_source_rows, updates = _prepare_source_rows_for_import(
+                        rows=rows,
+                        source_table=table,
+                        execution_attempt_ids=execution_attempt_ids,
+                    )
+                    deferred_source_preferred_attempt_updates.extend(updates)
+                    connection.execute(table.insert(), prepared_source_rows)
+                    continue
+                connection.execute(table.insert(), [_deserialize_row(row, table) for row in rows])
+
+            if deferred_source_preferred_attempt_updates:
+                source_table = SQLModel.metadata.tables["source"]
+                connection.execute(
+                    source_table.update()
+                    .where(source_table.c.id == bindparam("source_id"))
+                    .values(preferred_execution_attempt_id=bindparam("preferred_execution_attempt_id")),
+                    deferred_source_preferred_attempt_updates,
+                )
+    finally:
+        engine.dispose()
+
+
+def _collect_execution_attempt_ids(payload: dict[str, Any]) -> set[str]:
+    execution_attempt_rows = payload.get("tables", {}).get("execution_attempt", [])
+    return {_normalize_uuid_like(row.get("id")) for row in execution_attempt_rows if row.get("id") is not None}
+
+
+def _prepare_source_rows_for_import(
+    *,
+    rows: list[dict[str, Any]],
+    source_table: Table,
+    execution_attempt_ids: set[str],
+) -> tuple[list[dict[str, Any]], list[dict[str, Any]]]:
+    prepared_source_rows: list[dict[str, Any]] = []
+    updates: list[dict[str, Any]] = []
+    for row in rows:
+        source_row = _deserialize_row(row, source_table)
+        source_id = source_row.get("id")
+        preferred_attempt_id = source_row.get("preferred_execution_attempt_id")
+        if (
+            source_id is not None
+            and preferred_attempt_id is not None
+            and _normalize_uuid_like(preferred_attempt_id) in execution_attempt_ids
+        ):
+            updates.append(
+                {
+                    "source_id": source_id,
+                    "preferred_execution_attempt_id": preferred_attempt_id,
+                }
+            )
+        source_row["preferred_execution_attempt_id"] = None
+        prepared_source_rows.append(source_row)
+    return prepared_source_rows, updates
+
+
+def _ensure_sqlite_target_parent_exists(target_db_url: str) -> None:
+    parsed = make_url(target_db_url)
+    if not parsed.drivername.startswith("sqlite"):
+        return
+    database = parsed.database
+    if not database or database == ":memory:":
+        return
+    Path(database).parent.mkdir(parents=True, exist_ok=True)
+
+
+def _reset_sqlite_target_file(target_db_url: str) -> None:
+    parsed = make_url(target_db_url)
+    if not parsed.drivername.startswith("sqlite"):
+        return
+    database = parsed.database
+    if not database or database == ":memory:":
+        return
+    target = Path(database)
+    if target.exists():
+        target.unlink()
+
+
+def migrate_via_bundle(paths: MigrationPaths) -> None:
+    export_bundle(
+        source_db_url=paths.source_db_url,
+        source_upload_dir=paths.source_upload_dir,
+        bundle_dir=paths.bundle_dir,
+    )
+    import_bundle(
+        target_db_url=paths.target_db_url,
+        target_upload_dir=paths.target_upload_dir,
+        bundle_dir=paths.bundle_dir,
+    )
+
+
+@dataclass(frozen=True)
+class MigrationVerificationReport:
+    source_counts: dict[str, int]
+    target_counts: dict[str, int]
+    mismatched_tables: dict[str, dict[str, int]]
+    integrity_violations: dict[str, int]
+    success: bool
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "success": self.success,
+            "source_counts": self.source_counts,
+            "target_counts": self.target_counts,
+            "mismatched_tables": self.mismatched_tables,
+            "integrity_violations": self.integrity_violations,
+        }
+
+
+def sqlite_url_from_path(path: Path) -> str:
+    return URL.create(drivername="sqlite", database=str(path)).render_as_string(hide_password=False)
+
+
+def default_sync_db_url(settings: Settings | None = None) -> str:
+    runtime_settings = settings or get_settings()
+    return get_database_url(runtime_settings).replace("+aiosqlite", "").replace("+asyncpg", "").replace("+psycopg", "")
+
+
+def verify_migration(*, source_db_url: str, target_db_url: str) -> MigrationVerificationReport:
+    source_counts = _table_counts(source_db_url)
+    target_counts = _table_counts(target_db_url)
+    mismatched_tables = {
+        table_name: {"source": source_counts[table_name], "target": target_counts[table_name]}
+        for table_name in VERIFICATION_TABLES
+        if source_counts[table_name] != target_counts[table_name]
+    }
+    integrity_violations = _integrity_violations(target_db_url)
+    success = not mismatched_tables and all(count == 0 for count in integrity_violations.values())
+    return MigrationVerificationReport(
+        source_counts=source_counts,
+        target_counts=target_counts,
+        mismatched_tables=mismatched_tables,
+        integrity_violations=integrity_violations,
+        success=success,
+    )
+
+
+def _table_counts(db_url: str) -> dict[str, int]:
+    engine = create_engine(db_url)
+    try:
+        metadata = MetaData()
+        metadata.reflect(bind=engine)
+        counts: dict[str, int] = {}
+        with engine.connect() as connection:
+            for table_name in VERIFICATION_TABLES:
+                table = metadata.tables.get(table_name)
+                if table is None:
+                    counts[table_name] = 0
+                    continue
+                counts[table_name] = int(connection.execute(select(func.count()).select_from(table)).scalar_one())
+        return counts
+    finally:
+        engine.dispose()
+
+
+def _integrity_violations(db_url: str) -> dict[str, int]:
+    checks = {
+        "orphan_source_document": (
+            "select count(*) from source s left join document d on d.id = s.document_id where d.id is null"
+        ),
+        "orphan_job_document": (
+            "select count(*) from job j left join document d on d.id = j.document_id where d.id is null"
+        ),
+        "orphan_job_source_job": (
+            "select count(*) from job_source js left join job j on j.id = js.job_id where j.id is null"
+        ),
+        "orphan_job_source_source": (
+            "select count(*) from job_source js left join source s on s.id = js.source_id where s.id is null"
+        ),
+        "orphan_attempt_job_source": (
+            "select count(*) from execution_attempt ea "
+            "left join job_source js on js.id = ea.job_source_id "
+            "where js.id is null"
+        ),
+        "orphan_attempt_job": (
+            "select count(*) from execution_attempt ea left join job j on j.id = ea.job_id where j.id is null"
+        ),
+        "orphan_attempt_source": (
+            "select count(*) from execution_attempt ea left join source s on s.id = ea.source_id where s.id is null"
+        ),
+        "duplicate_attempt_numbers": (
+            "select count(*) from ("
+            "  select job_id, source_id, attempt_number, count(*) as c"
+            "  from execution_attempt"
+            "  group by job_id, source_id, attempt_number"
+            "  having count(*) > 1"
+            ") x"
+        ),
+    }
+    engine = create_engine(db_url)
+    try:
+        with engine.connect() as connection:
+            return {
+                check_name: int(connection.execute(text(query)).scalar_one()) for check_name, query in checks.items()
+            }
+    finally:
+        engine.dispose()
+
+
+def _serialize_row(row: RowMapping, *, table_name: str, source_upload_dir: Path) -> dict[str, Any]:
+    serialized: dict[str, Any] = {}
+    for raw_key, value in row.items():
+        key = str(raw_key)
+        serialized_value = _serialize_value(value)
+        if table_name == "source" and key == "file_path" and isinstance(serialized_value, str):
+            serialized[key] = _canonical_media_relative_path(
+                serialized_value,
+                source_upload_dir=source_upload_dir,
+                preferred_prefix="documents/",
+            )
+            continue
+        if table_name == "photo" and key == "path" and isinstance(serialized_value, str):
+            serialized[key] = _canonical_media_relative_path(
+                serialized_value,
+                source_upload_dir=source_upload_dir,
+                preferred_prefix="photos/",
+            )
+            continue
+        if table_name == "person" and key == "full_name" and isinstance(serialized_value, str):
+            given_names, last_name = _split_legacy_full_name(serialized_value)
+            serialized["given_names"] = given_names
+            serialized["last_name"] = last_name
+            continue
+        serialized[key] = serialized_value
+    if table_name == "person":
+        serialized["given_names"] = str(serialized.get("given_names") or "").strip()
+        serialized["last_name"] = str(serialized.get("last_name") or "").strip()
+    return serialized
+
+
+def _split_legacy_full_name(full_name: str) -> tuple[str, str]:
+    tokens = [token for token in full_name.strip().split() if token]
+    if len(tokens) >= 2:
+        return (" ".join(tokens[:-1]), tokens[-1])
+    if len(tokens) == 1:
+        return (tokens[0], tokens[0])
+    return ("Unknown", "Unknown")
+
+
+def _serialize_value(value: Any) -> Any:
+    if isinstance(value, UUID):
+        return str(value)
+    if isinstance(value, (datetime, date)):
+        return value.isoformat()
+    if isinstance(value, bytes):
+        return {"encoding": "base64", "data": base64.b64encode(value).decode("ascii")}
+    if isinstance(value, dict):
+        return {str(k): _serialize_value(v) for k, v in value.items()}
+    if isinstance(value, list):
+        return [_serialize_value(item) for item in value]
+    return value
+
+
+def _deserialize_row(row: dict[str, Any], table: Table) -> dict[str, Any]:
+    deserialized: dict[str, Any] = {}
+    for key, value in row.items():
+        if key in BYTES_FIELDS and isinstance(value, dict) and value.get("encoding") == "base64":
+            deserialized[key] = base64.b64decode(value["data"])
+            continue
+        if key in table.columns:
+            try:
+                python_type: type[Any] = table.columns[key].type.python_type
+            except NotImplementedError:
+                deserialized[key] = value
+                continue
+            deserialized[key] = _deserialize_value(python_type, value)
+    return deserialized
+
+
+def _deserialize_value(python_type: type[Any], value: Any) -> Any:
+    if value is None:
+        return None
+    if python_type is UUID and isinstance(value, str):
+        return UUID(value)
+    if python_type is datetime and isinstance(value, str):
+        return datetime.fromisoformat(value)
+    if python_type is date and isinstance(value, str):
+        return date.fromisoformat(value)
+    return value
+
+
+def _normalize_uuid_like(value: Any) -> str:
+    if isinstance(value, UUID):
+        return str(value)
+    if isinstance(value, str):
+        text_value = value.strip()
+        try:
+            return str(UUID(text_value))
+        except ValueError:
+            return text_value
+    return str(value)
+
+
+def _canonical_media_relative_path(value: str, *, source_upload_dir: Path, preferred_prefix: str) -> str:
+    normalized = value.strip().replace("\\", "/")
+    lowered = normalized.casefold()
+    upload_root = source_upload_dir.resolve().as_posix().casefold().rstrip("/")
+    if lowered.startswith(upload_root + "/"):
+        normalized = normalized[len(source_upload_dir.resolve().as_posix()) + 1 :]
+        lowered = normalized.casefold()
+
+    if lowered.startswith("/uploads/"):
+        normalized = normalized[len("/uploads/") :]
+        lowered = normalized.casefold()
+    elif lowered.startswith("uploads/"):
+        normalized = normalized[len("uploads/") :]
+        lowered = normalized.casefold()
+    elif lowered.startswith("data/"):
+        normalized = normalized[len("data/") :]
+        lowered = normalized.casefold()
+
+    if preferred_prefix == "persons/" and lowered.startswith("portraits/"):
+        normalized = "persons/" + normalized[len("portraits/") :]
+        lowered = normalized.casefold()
+
+    for prefix in ("documents/", "photos/", "persons/"):
+        marker = f"/{prefix}"
+        index = lowered.find(marker)
+        if index >= 0:
+            normalized = normalized[index + 1 :]
+            lowered = normalized.casefold()
+            break
+
+    if not lowered.startswith(preferred_prefix):
+        return normalized
+    return Path(normalized).as_posix()
+
+
+def _prepare_photo_payload_and_uploads(  # noqa: PLR0915
+    *,
+    payload: dict[str, Any],
+    uploads_bundle_dir: Path,
+    legacy_portrait_rows: Sequence[RowMapping],
+) -> None:
+    photo_rows = payload.setdefault("tables", {}).setdefault("photo", [])
+    photos_dir = uploads_bundle_dir / "photos"
+    photos_dir.mkdir(parents=True, exist_ok=True)
+
+    # Keep only photo rows whose referenced media exists inside the uploads tree.
+    # This prevents stale/injected rows from blocking legacy backfill.
+    retained_rows: list[dict[str, Any]] = []
+    for row in photo_rows:
+        path_value = row.get("path")
+        if not isinstance(path_value, str) or not path_value.strip():
+            continue
+        canonical_path = _canonical_media_relative_path(
+            path_value,
+            source_upload_dir=uploads_bundle_dir,
+            preferred_prefix="photos/",
+        )
+        candidate = uploads_bundle_dir / canonical_path
+        if not candidate.exists():
+            continue
+        row["path"] = canonical_path
+        retained_rows.append(row)
+    photo_rows[:] = retained_rows
+
+    existing_homepage_rows = [row for row in photo_rows if row.get("person_id") is None]
+    existing_person_ids = {str(row["person_id"]) for row in photo_rows if row.get("person_id") is not None}
+    existing_primary_person_ids = {
+        str(row["person_id"]) for row in photo_rows if row.get("person_id") is not None and bool(row.get("is_primary"))
+    }
+    has_homepage_primary = any(bool(row.get("is_primary")) for row in existing_homepage_rows)
+
+    now_iso = datetime.now(UTC).isoformat()
+    for row in legacy_portrait_rows:
+        portrait_path = row.get("portrait_path")
+        person_id = row.get("id")
+        if not isinstance(portrait_path, str) or not portrait_path.strip():
+            continue
+        if person_id is None:
+            continue
+        canonical = _canonical_media_relative_path(
+            portrait_path,
+            source_upload_dir=uploads_bundle_dir,
+            preferred_prefix="persons/",
+        )
+        source_file = uploads_bundle_dir / canonical
+        if not source_file.exists():
+            continue
+        person_key = str(person_id)
+        if person_key in existing_person_ids:
+            continue
+        suffix = Path(canonical).suffix.lower() or ".jpg"
+        photo_id = str(uuid4())
+        relative_path = f"photos/{photo_id}{suffix}"
+        target_file = uploads_bundle_dir / relative_path
+        target_file.parent.mkdir(parents=True, exist_ok=True)
+        shutil.copy2(source_file, target_file)
+        is_primary = person_key not in existing_primary_person_ids
+        photo_rows.append(
+            {
+                "id": photo_id,
+                "person_id": person_key,
+                "path": relative_path,
+                "description": None,
+                "is_primary": is_primary,
+                "created_at": now_iso,
+                "updated_at": now_iso,
+            }
+        )
+        existing_person_ids.add(person_key)
+        if is_primary:
+            existing_primary_person_ids.add(person_key)
+
+    legacy_homepage_dir = uploads_bundle_dir / "homepage"
+    if not legacy_homepage_dir.exists():
+        return
+
+    homepage_images = sorted(
+        [
+            path
+            for path in legacy_homepage_dir.iterdir()
+            if path.is_file()
+            and path.suffix.lower() in {".jpg", ".jpeg", ".png", ".gif", ".webp", ".bmp", ".tif", ".tiff"}
+        ],
+        key=lambda path: (path.stat().st_mtime, path.name),
+    )
+    if existing_homepage_rows:
+        return
+    for index, image_path in enumerate(homepage_images):
+        photo_id = str(uuid4())
+        relative_path = f"photos/{photo_id}{image_path.suffix.lower()}"
+        target_file = uploads_bundle_dir / relative_path
+        target_file.parent.mkdir(parents=True, exist_ok=True)
+        shutil.copy2(image_path, target_file)
+        photo_rows.append(
+            {
+                "id": photo_id,
+                "person_id": None,
+                "path": relative_path,
+                "description": None,
+                "is_primary": (not has_homepage_primary) and index == 0,
+                "created_at": now_iso,
+                "updated_at": now_iso,
+            }
+        )
+
+
+def _relocate_homepage_markdown(*, uploads_bundle_dir: Path) -> None:
+    legacy_markdown = uploads_bundle_dir / "homepage" / "homepage.md"
+    target_markdown = uploads_bundle_dir / "homepage.md"
+    if not legacy_markdown.exists() or target_markdown.exists():
+        return
+    target_markdown.parent.mkdir(parents=True, exist_ok=True)
+    shutil.copy2(legacy_markdown, target_markdown)

src/transcription/db/models.py

@@ -0,0 +1,769 @@
+"""SQLModel domain models for the V3 transcription system."""
+
+from datetime import UTC
+from datetime import date
+from datetime import datetime
+from enum import StrEnum
+from typing import Any
+from typing import Optional
+from uuid import UUID
+from uuid import uuid4
+
+from pydantic import JsonValue
+from sqlalchemy import JSON
+from sqlalchemy import BigInteger
+from sqlalchemy import Column
+from sqlalchemy import Enum as SAEnum
+from sqlalchemy import ForeignKey
+from sqlalchemy import Index
+from sqlalchemy import LargeBinary
+from sqlalchemy import UniqueConstraint
+from sqlalchemy import Uuid
+from sqlalchemy import inspect as sqlalchemy_inspect
+from sqlalchemy.dialects.postgresql import JSONB
+from sqlalchemy.exc import NoInspectionAvailable
+from sqlalchemy.orm.state import InstanceState
+from sqlalchemy.types import TypeDecorator
+from sqlmodel import Field
+from sqlmodel import Relationship
+from sqlmodel import SQLModel
+
+
+def _loaded_attribute(instance: object, attribute: str) -> Any | None:
+    """Return ``attribute`` only when it is already loaded on ``instance``.
+
+    Relationships in this module declare ``lazy="raise"``, so reading an
+    unloaded attribute is an error rather than a silent query. Callers that
+    render optional detail use this to distinguish "not loaded" from "absent"
+    without catching exceptions indiscriminately.
+    """
+    try:
+        state: InstanceState[Any] = sqlalchemy_inspect(instance, raiseerr=True)
+    except NoInspectionAvailable:
+        return None
+    if attribute in state.unloaded:
+        return None
+    return state.dict.get(attribute)
+
+
+def _utc_now_naive() -> datetime:
+    """Return a UTC timestamp stored as a naive datetime."""
+    return datetime.now(UTC).replace(tzinfo=None)
+
+
+class JSONBCompat(TypeDecorator):
+    """JSONB for PostgreSQL and JSON for SQLite/testing backends."""
+
+    impl = JSON(none_as_null=True)
+
+    def load_dialect_impl(self, dialect):
+        if dialect.name == "postgresql":
+            return dialect.type_descriptor(JSONB(none_as_null=True))
+        return dialect.type_descriptor(JSON(none_as_null=True))
+
+
+class JobStatus(StrEnum):
+    QUEUED = "queued"
+    PROCESSING = "processing"
+    TRANSCRIBED = "transcribed"
+    PARTIAL_SUCCESS = "partial_success"
+    FAILED = "failed"
+
+
+class JobSourceStatus(StrEnum):
+    PENDING = "pending"
+    TRANSCRIBED = "transcribed"
+    FAILED = "failed"
+    CANCELLED = "cancelled"
+
+
+class JobPurpose(StrEnum):
+    TRANSCRIPTION = "transcription"
+    RETRANSCRIPTION = "retranscription"
+
+
+class MaintenanceJobType(StrEnum):
+    BACKUP = "backup"
+    STORAGE_RECONCILIATION = "storage_reconciliation"
+    GEDCOM_IMPORT = "gedcom_import"
+
+
+class MaintenanceRunStatus(StrEnum):
+    QUEUED = "queued"
+    PROCESSING = "processing"
+    SUCCEEDED = "succeeded"
+    FAILED = "failed"
+
+
+class GenealogyCitationFactType(StrEnum):
+    BIRTH = "birth"
+    DEATH = "death"
+    MARRIAGE = "marriage"
+    OTHER = "other"
+
+
+class GenealogyCitationSourceKind(StrEnum):
+    FAMILYSEARCH_IMPORTED = "familysearch_imported"
+    TRANSCRIPTION_EVIDENCE = "transcription_evidence"
+
+
+class DocumentType(SQLModel, table=True):
+    """Registry of allowed document types."""
+
+    __tablename__ = "document_type"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    semantic_key: str | None = Field(default=None, index=True, unique=True)
+    label: str
+    normalized_label: str = Field(index=True, unique=True)
+    is_active: bool = True
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    documents: list["Document"] = Relationship(
+        back_populates="document_type_ref", sa_relationship_kwargs={"lazy": "raise"}
+    )
+
+
+class PersonRole(SQLModel, table=True):
+    """Registry of allowed document-person relationship roles."""
+
+    __tablename__ = "person_role"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    semantic_key: str | None = Field(default=None, index=True, unique=True)
+    label: str
+    normalized_label: str = Field(index=True, unique=True)
+    is_active: bool = True
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    document_people: list["DocumentPerson"] = Relationship(
+        back_populates="role_ref", sa_relationship_kwargs={"lazy": "raise"}
+    )
+
+
+class Tag(SQLModel, table=True):
+    """Registry of labels that can be attached to Documents."""
+
+    __tablename__ = "tag"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    semantic_key: str | None = Field(default=None, index=True, unique=True)
+    label: str
+    normalized_label: str = Field(index=True, unique=True)
+    is_active: bool = True
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    document_tags: list["DocumentTag"] = Relationship(
+        back_populates="tag_ref",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    person_tags: list["PersonTag"] = Relationship(
+        back_populates="tag_ref",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class Document(SQLModel, table=True):
+    """An historical document."""
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    name: str
+    document_type_id: UUID | None = Field(default=None, foreign_key="document_type.id", index=True)
+    document_date: date | None = None
+    document_date_raw: str | None = None
+    location_created: str | None = None
+    notes: str | None = None
+    archive_identifier: str | None = None
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    jobs: list["Job"] = Relationship(back_populates="document", sa_relationship_kwargs={"lazy": "raise"})
+    sources: list["Source"] = Relationship(back_populates="document", sa_relationship_kwargs={"lazy": "raise"})
+    document_people: list["DocumentPerson"] = Relationship(
+        back_populates="document", sa_relationship_kwargs={"lazy": "raise"}
+    )
+    document_tags: list["DocumentTag"] = Relationship(
+        back_populates="document",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    document_type_ref: Optional["DocumentType"] = Relationship(
+        back_populates="documents", sa_relationship_kwargs={"lazy": "raise"}
+    )
+
+
+class Person(SQLModel, table=True):
+    """A historical person linked to one or more documents."""
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    last_name: str
+    given_names: str
+    birth_date: date | None = None
+    birth_date_raw: str | None = None
+    birth_place: str | None = None
+    death_date: date | None = None
+    death_date_raw: str | None = None
+    death_place: str | None = None
+    biography: str | None = None
+    family_search_id: str | None = Field(default=None, unique=True)
+    metadata_: dict[str, JsonValue] | None = Field(
+        default=None,
+        sa_column=Column("metadata", JSONBCompat(), nullable=True),
+    )
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    document_people: list["DocumentPerson"] = Relationship(
+        back_populates="person", sa_relationship_kwargs={"lazy": "raise"}
+    )
+    person_tags: list["PersonTag"] = Relationship(
+        back_populates="person",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    photos: list["Photo"] = Relationship(
+        back_populates="person",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+    @property
+    def full_name(self) -> str:
+        """Presentation-friendly combined name."""
+        return f"{self.given_names} {self.last_name}".strip()
+
+
+class GenealogyPerson(SQLModel, table=True):
+    """An individual imported from a GEDCOM export."""
+
+    __tablename__ = "genealogy_person"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    fs_id: str = Field(index=True, unique=True)
+    full_name: str
+    birth_date: date | None = None
+    birth_date_raw: str | None = None
+    birth_place: str | None = None
+    death_date: date | None = None
+    death_date_raw: str | None = None
+    death_place: str | None = None
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    husband_families: list["GenealogyFamily"] = Relationship(
+        back_populates="husband",
+        sa_relationship_kwargs={"lazy": "raise", "foreign_keys": "[GenealogyFamily.husband_id]"},
+    )
+    wife_families: list["GenealogyFamily"] = Relationship(
+        back_populates="wife",
+        sa_relationship_kwargs={"lazy": "raise", "foreign_keys": "[GenealogyFamily.wife_id]"},
+    )
+    child_family_memberships: list["GenealogyFamilyChild"] = Relationship(
+        back_populates="child",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    citations: list["GenealogyCitation"] = Relationship(
+        back_populates="genealogy_person",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class GenealogyFamily(SQLModel, table=True):
+    """A family linking two GenealogyPerson records."""
+
+    __tablename__ = "genealogy_family"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    fs_family_id: str = Field(index=True, unique=True)
+    husband_id: UUID | None = Field(default=None, foreign_key="genealogy_person.id", index=True)
+    wife_id: UUID | None = Field(default=None, foreign_key="genealogy_person.id", index=True)
+    marriage_date: date | None = None
+    marriage_date_raw: str | None = None
+    marriage_place: str | None = None
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    husband: Optional["GenealogyPerson"] = Relationship(
+        back_populates="husband_families",
+        sa_relationship_kwargs={"lazy": "raise", "foreign_keys": "[GenealogyFamily.husband_id]"},
+    )
+    wife: Optional["GenealogyPerson"] = Relationship(
+        back_populates="wife_families",
+        sa_relationship_kwargs={"lazy": "raise", "foreign_keys": "[GenealogyFamily.wife_id]"},
+    )
+    children: list["GenealogyFamilyChild"] = Relationship(
+        back_populates="family",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    citations: list["GenealogyCitation"] = Relationship(
+        back_populates="genealogy_family",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class GenealogyFamilyChild(SQLModel, table=True):
+    """Junction table for child membership within a genealogy family."""
+
+    __tablename__ = "genealogy_family_child"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    family_id: UUID = Field(foreign_key="genealogy_family.id", index=True)
+    child_id: UUID = Field(foreign_key="genealogy_person.id", index=True)
+    relationship_type: str | None = None
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+
+    __table_args__ = (UniqueConstraint("family_id", "child_id", name="uq_genealogy_family_child"),)
+
+    family: Optional["GenealogyFamily"] = Relationship(
+        back_populates="children",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    child: Optional["GenealogyPerson"] = Relationship(
+        back_populates="child_family_memberships",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class GenealogyCitation(SQLModel, table=True):
+    """A source citation attached to a genealogical fact."""
+
+    __tablename__ = "genealogy_citation"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    genealogy_person_id: UUID | None = Field(default=None, foreign_key="genealogy_person.id", index=True)
+    genealogy_family_id: UUID | None = Field(default=None, foreign_key="genealogy_family.id", index=True)
+    fact_type: GenealogyCitationFactType = Field(
+        sa_column=Column(
+            SAEnum(
+                GenealogyCitationFactType,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+        )
+    )
+    raw_citation_text: str
+    source_kind: GenealogyCitationSourceKind = Field(
+        sa_column=Column(
+            SAEnum(
+                GenealogyCitationSourceKind,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+        )
+    )
+    document_id: UUID | None = Field(default=None, foreign_key="document.id", index=True)
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+
+    genealogy_person: Optional["GenealogyPerson"] = Relationship(
+        back_populates="citations",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    genealogy_family: Optional["GenealogyFamily"] = Relationship(
+        back_populates="citations",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    document: Optional["Document"] = Relationship(sa_relationship_kwargs={"lazy": "raise"})
+
+
+class Photo(SQLModel, table=True):
+    """A reusable image record for Person and homepage galleries."""
+
+    __tablename__ = "photo"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    person_id: UUID | None = Field(default=None, foreign_key="person.id", index=True)
+    path: str
+    description: str | None = None
+    is_primary: bool = False
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    person: Optional["Person"] = Relationship(
+        back_populates="photos",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class DocumentPerson(SQLModel, table=True):
+    """Associates documents with people in a given role."""
+
+    __tablename__ = "document_person"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    document_id: UUID = Field(foreign_key="document.id", index=True)
+    person_id: UUID = Field(foreign_key="person.id", index=True)
+    role_id: UUID = Field(foreign_key="person_role.id", index=True)
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    __table_args__ = (UniqueConstraint("document_id", "person_id", name="uq_document_person"),)
+
+    document: Optional["Document"] = Relationship(
+        back_populates="document_people", sa_relationship_kwargs={"lazy": "raise"}
+    )
+    person: Optional["Person"] = Relationship(
+        back_populates="document_people", sa_relationship_kwargs={"lazy": "raise"}
+    )
+    role_ref: Optional["PersonRole"] = Relationship(
+        back_populates="document_people", sa_relationship_kwargs={"lazy": "raise"}
+    )
+
+
+class DocumentTag(SQLModel, table=True):
+    """Associates Documents with Tags."""
+
+    __tablename__ = "document_tag"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    document_id: UUID = Field(foreign_key="document.id", index=True)
+    tag_id: UUID = Field(foreign_key="tag.id", index=True)
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    __table_args__ = (UniqueConstraint("document_id", "tag_id", name="uq_document_tag"),)
+
+    document: Optional["Document"] = Relationship(
+        back_populates="document_tags",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    tag_ref: Optional["Tag"] = Relationship(
+        back_populates="document_tags",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class PersonTag(SQLModel, table=True):
+    """Associates People with Tags."""
+
+    __tablename__ = "person_tag"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    person_id: UUID = Field(foreign_key="person.id", index=True)
+    tag_id: UUID = Field(foreign_key="tag.id", index=True)
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+    __table_args__ = (UniqueConstraint("person_id", "tag_id", name="uq_person_tag"),)
+
+    person: Optional["Person"] = Relationship(
+        back_populates="person_tags",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    tag_ref: Optional["Tag"] = Relationship(
+        back_populates="person_tags",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+
+class Job(SQLModel, table=True):
+    """A transcription job tied to a single document."""
+
+    __table_args__ = (Index("ix_job_status_date_created", "status", "date_created"),)
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    document_id: UUID = Field(foreign_key="document.id", index=True)
+    status: JobStatus = Field(
+        default=JobStatus.QUEUED,
+        sa_column=Column(
+            SAEnum(
+                JobStatus,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+        ),
+    )
+    retry_count: int = Field(default=0, ge=0)
+    purpose: JobPurpose = Field(
+        default=JobPurpose.TRANSCRIPTION,
+        sa_column=Column(
+            SAEnum(
+                JobPurpose,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+            default=JobPurpose.TRANSCRIPTION.value,
+        ),
+    )
+    date_created: datetime = Field(default_factory=_utc_now_naive)
+    date_updated: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+    provider: str | None = None
+    model: str | None = None
+    prompt_name: str | None = None
+    prompt_hash: str | None = None
+    system_prompt: str | None = None
+    user_prompt: str | None = None
+    temperature: float | None = None
+    top_p: float | None = None
+
+    document: Optional["Document"] = Relationship(back_populates="jobs", sa_relationship_kwargs={"lazy": "raise"})
+    job_sources: list["JobSource"] = Relationship(back_populates="job", sa_relationship_kwargs={"lazy": "raise"})
+
+    @property
+    def filename(self) -> str:
+        """Return the filename of the first loaded source, when available.
+
+        Relationships on this model use ``lazy="raise"``, so this deliberately
+        inspects load state rather than triggering (or swallowing) a lazy load:
+        a read model that did not eager-load its sources gets "unknown" instead
+        of an unhandled error, and genuine errors are no longer hidden.
+        """
+        for job_source in _loaded_attribute(self, "job_sources") or ():
+            source = _loaded_attribute(job_source, "source")
+            if source is not None:
+                return source.filename
+
+        return "unknown"
+
+
+class MaintenanceRun(SQLModel, table=True):
+    """A queued/processed maintenance task execution record."""
+
+    __tablename__ = "maintenance_run"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    job_type: MaintenanceJobType = Field(
+        sa_column=Column(
+            SAEnum(
+                MaintenanceJobType,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+        )
+    )
+    status: MaintenanceRunStatus = Field(
+        default=MaintenanceRunStatus.QUEUED,
+        sa_column=Column(
+            SAEnum(
+                MaintenanceRunStatus,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+            default=MaintenanceRunStatus.QUEUED.value,
+        ),
+    )
+    started_at: datetime | None = None
+    finished_at: datetime | None = None
+    triggered_by: str | None = None
+    summary: str | None = None
+    log_path: str | None = None
+    error_detail: str | None = None
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+    updated_at: datetime = Field(
+        default_factory=_utc_now_naive,
+        sa_column_kwargs={"onupdate": _utc_now_naive},
+    )
+
+
+class Source(SQLModel, table=True):
+    """A document source image or PDF page."""
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    document_id: UUID = Field(foreign_key="document.id", index=True)
+    page_number: int = Field(default=1, ge=1)
+    upload_name: str
+    filename: str
+    file_path: str
+    file_hash: str
+    file_size_bytes: int = Field(sa_column=Column(BigInteger(), nullable=False))
+    raw_transcription: str | None = None
+    preferred_execution_attempt_id: UUID | None = Field(
+        default=None,
+        sa_column=Column(
+            Uuid(),
+            # use_alter breaks the source / job_source / execution_attempt cycle so
+            # metadata.create_all can order table creation on every dialect.
+            ForeignKey(
+                "execution_attempt.id",
+                use_alter=True,
+                name="fk_source_preferred_execution_attempt_id",
+            ),
+            nullable=True,
+            index=True,
+        ),
+    )
+    revised_text: str | None = None
+    date_uploaded: datetime = Field(default_factory=_utc_now_naive)
+    date_revised: datetime | None = None
+
+    document: Optional["Document"] = Relationship(
+        back_populates="sources",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+    job_sources: list["JobSource"] = Relationship(
+        back_populates="source",
+        sa_relationship_kwargs={"lazy": "raise"},
+    )
+
+    @property
+    def latest_job_source(self) -> Optional["JobSource"]:
+        """Return the most recent job execution record for this source.
+
+        ``JobSource`` carries no timestamp of its own, so recency is the parent
+        job's creation time. ``(job_id, source_id)`` is unique per source, so
+        this is exactly "the most recent job that included this page".
+        """
+        job_sources = _loaded_attribute(self, "job_sources") or ()
+        dated = [
+            (job, job_source) for job_source in job_sources if (job := _loaded_attribute(job_source, "job")) is not None
+        ]
+        if dated:
+            return max(dated, key=lambda pair: pair[0].date_created)[1]
+        return job_sources[0] if job_sources else None
+
+    @property
+    def latest_status(self) -> JobSourceStatus | None:
+        """Return the execution status of the latest job run."""
+        latest = self.latest_job_source
+        return latest.status if latest else None
+
+    @property
+    def latest_error_detail(self) -> str | None:
+        """Return the error detail of the latest attempt on the latest job run.
+
+        Failure detail lives on ``ExecutionAttempt``; ``JobSource`` records only
+        which page a job is working on and how far it got.
+        """
+        latest = self.latest_job_source
+        if latest is None:
+            return None
+        attempts = _loaded_attribute(latest, "execution_attempts") or ()
+        if not attempts:
+            return None
+        latest_attempt = max(attempts, key=lambda item: item.attempt_number)
+        return latest_attempt.error_detail
+
+    @property
+    def document_name(self) -> str | None:
+        """Return the parent document name if loaded."""
+        return self.document.name if self.document else None
+
+
+class JobSource(SQLModel, table=True):
+    """A single AI execution record for one source page."""
+
+    __tablename__ = "job_source"
+    __table_args__ = (UniqueConstraint("job_id", "source_id", name="uq_job_source_job_source"),)
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    job_id: UUID = Field(foreign_key="job.id", index=True)
+    source_id: UUID = Field(foreign_key="source.id", index=True)
+    status: JobSourceStatus = Field(
+        default=JobSourceStatus.PENDING,
+        sa_column=Column(
+            SAEnum(
+                JobSourceStatus,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+        ),
+    )
+
+    job: Optional["Job"] = Relationship(back_populates="job_sources", sa_relationship_kwargs={"lazy": "raise"})
+    source: Optional["Source"] = Relationship(back_populates="job_sources", sa_relationship_kwargs={"lazy": "raise"})
+    execution_attempts: list["ExecutionAttempt"] = Relationship(
+        back_populates="job_source",
+        sa_relationship_kwargs={"lazy": "noload", "order_by": "ExecutionAttempt.attempt_number"},
+    )
+
+
+class ExecutionAttempt(SQLModel, table=True):
+    """Immutable evidence for one provider call attempt."""
+
+    __tablename__ = "execution_attempt"
+    __table_args__ = (UniqueConstraint("job_id", "source_id", "attempt_number", name="uq_execution_attempt_number"),)
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    job_source_id: UUID = Field(foreign_key="job_source.id", index=True)
+    job_id: UUID = Field(foreign_key="job.id", index=True)
+    source_id: UUID = Field(foreign_key="source.id", index=True)
+    attempt_number: int = Field(ge=1)
+    status: JobSourceStatus = Field(
+        sa_column=Column(
+            # Declared identically to job_source.status. Without values_callable
+            # SQLAlchemy persists enum *names*, which is defect [45]: the two
+            # columns spelled the same status differently and never compared equal.
+            SAEnum(
+                JobSourceStatus,
+                values_callable=lambda enum_cls: [item.value for item in enum_cls],
+                native_enum=False,
+            ),
+            nullable=False,
+        )
+    )
+    provider: str
+    model: str | None = None
+    request_manifest: dict[str, JsonValue] | None = Field(default=None, sa_column=Column(JSONBCompat(), nullable=True))
+    request_manifest_sha256: str | None = None
+    request_manifest_schema_version: str | None = None
+    response_received: bool = False
+    transport_status_code: int | None = None
+    transport_body: bytes | None = Field(default=None, sa_column=Column(LargeBinary(), nullable=True))
+    transport_content_type: str | None = None
+    transport_content_encoding: str | None = None
+    transport_safe_headers: dict[str, JsonValue] | None = Field(
+        default=None, sa_column=Column(JSONBCompat(), nullable=True)
+    )
+    router_request_id: str | None = None
+    router_generation_id: str | None = None
+    sdk_response_snapshot: dict[str, JsonValue] | None = Field(
+        default=None, sa_column=Column(JSONBCompat(), nullable=True)
+    )
+    normalized_metadata: dict[str, JsonValue] | None = Field(
+        default=None, sa_column=Column(JSONBCompat(), nullable=True)
+    )
+    software_context: dict[str, JsonValue] | None = Field(default=None, sa_column=Column(JSONBCompat(), nullable=True))
+    raw_transcription: str | None = None
+    error_category: str | None = None
+    error_detail: str | None = None
+    failure_phase: str | None = None
+    started_at: datetime
+    finished_at: datetime
+    duration_ms: int = Field(ge=0)
+    created_at: datetime = Field(default_factory=_utc_now_naive)
+
+    job_source: Optional["JobSource"] = Relationship(
+        back_populates="execution_attempts", sa_relationship_kwargs={"lazy": "raise"}
+    )

src/transcription/db/operations.py

@@ -1,79 +1,248 @@
 from __future__ import annotations
 
 import logging
+from pathlib import Path
 
-from sqlalchemy import inspect
+from sqlalchemy import inspect as sqlalchemy_inspect
 from sqlalchemy import text
-from sqlalchemy.engine import Connection
 from sqlalchemy.ext.asyncio import AsyncEngine
+from sqlalchemy.ext.asyncio import async_sessionmaker
 from sqlmodel import SQLModel
 from sqlmodel import select
 from sqlmodel.ext.asyncio.session import AsyncSession
 
-from ..models import Job
-from ..models import JobStatus
-from .runtime import get_engine
+from .engine import resolve_engine
+from .models import DocumentType
+from .models import PersonRole
+from .registries import BUILT_IN_DOCUMENT_TYPES
+from .registries import BUILT_IN_PERSON_ROLES
 
 logger = logging.getLogger(__name__)
 
 
-async def get_next_queued_job(*, session: AsyncSession) -> Job | None:
-    """Get the next queued job, if any."""
-    result = await session.exec(
-        select(Job)
-        .where(Job.status == JobStatus.QUEUED)
-        .order_by(Job.date_created)  # pyright: ignore[reportArgumentType]
-        .limit(1)
-    )  # fmt: skip
-    return result.first()
-
-
 async def create_all(*, engine: AsyncEngine | None = None) -> None:
-    """Create all tables on the selected engine."""
+    """Create any missing tables on the selected engine."""
     # Import models so SQLModel metadata is fully registered before bootstrap.
-    from transcription import models as _models  # noqa: F401
+    from transcription.db import models as _models  # noqa: F401
 
-    active_engine = engine or get_engine()
+    active_engine = engine or resolve_engine()
     async with active_engine.begin() as connection:
         await connection.run_sync(SQLModel.metadata.create_all)
-        await connection.run_sync(_ensure_sqlite_compat_columns)
+    await seed_registry_defaults(engine=active_engine)
     logger.debug("Database schema bootstrap complete for database_url=%s", active_engine.url)
 
 
-def _ensure_sqlite_compat_columns(connection: Connection) -> None:
-    """Apply lightweight dev/test SQLite compatibility column patches.
+async def reconcile_legacy_job_source_columns(*, engine: AsyncEngine | None = None) -> int:
+    """Remove stale V4.6 ``job_source`` evidence columns from existing databases.
 
-    This keeps local bootstrap resilient when models evolve but no full
-    migration tooling is in place yet.
+    Runtime models define ``job_source`` as a queue/projection table only. If an
+    older database still carries the retired evidence columns, writes can fail
+    on stale constraints (for example ``executed_at NOT NULL``).
     """
-    if connection.engine.url.get_backend_name() != "sqlite":
-        return
+    active_engine = engine or resolve_engine()
+    if not hasattr(active_engine, "begin"):
+        return 0
 
-    inspector = inspect(connection)
+    def _reconcile(sync_connection) -> int:
+        inspector = sqlalchemy_inspect(sync_connection)
         table_names = set(inspector.get_table_names())
+        if "job_source" not in table_names:
+            return 0
+        present_columns = {column["name"] for column in inspector.get_columns("job_source")}
+        dropped = 0
+        for column_name in (
+            "raw_transcription",
+            "ai_metadata",
+            "raw_api_response",
+            "error_detail",
+            "executed_at",
+        ):
+            if column_name not in present_columns:
+                continue
+            sync_connection.execute(text(f'alter table "job_source" drop column "{column_name}"'))
+            dropped += 1
+        return dropped
 
-    if "job" in table_names:
-        job_columns = {column["name"] for column in inspector.get_columns("job")}
-        if "retry_count" not in job_columns:
-            connection.execute(text("ALTER TABLE job ADD COLUMN retry_count INTEGER NOT NULL DEFAULT 0"))
-            logger.warning("Applied SQLite compatibility schema patch table=job column=retry_count default=0")
+    async with active_engine.begin() as connection:
+        dropped_columns = await connection.run_sync(_reconcile)
+    if dropped_columns:
+        logger.warning("Dropped %s legacy job_source column(s) during startup reconciliation", dropped_columns)
+    return dropped_columns
 
-    if "revision" in table_names:
-        revision_columns = {column["name"] for column in inspector.get_columns("revision")}
-        if "source_id" in revision_columns:
-            has_unique_source = False
-            for index in inspector.get_indexes("revision"):
-                if index.get("unique") and index.get("column_names") == ["source_id"]:
-                    has_unique_source = True
+
+async def reconcile_canonical_media_paths(*, engine: AsyncEngine | None = None) -> int:
+    """Normalize stored media paths to upload-root-relative POSIX form."""
+    active_engine = engine or resolve_engine()
+    if not hasattr(active_engine, "begin"):
+        return 0
+
+    def _reconcile(sync_connection) -> int:
+        rows_changed = 0
+        inspector = sqlalchemy_inspect(sync_connection)
+        table_names = set(inspector.get_table_names())
+        if "source" in table_names:
+            rows = (
+                sync_connection.execute(text('select id, file_path from "source" where file_path is not null'))
+                .mappings()
+                .all()
+            )
+            for row in rows:
+                original = str(row["file_path"])
+                normalized = _canonical_relative_path(original, preferred_prefix="documents/")
+                if normalized is None or normalized == original:
+                    continue
+                sync_connection.execute(
+                    text('update "source" set file_path = :file_path where id = :id'),
+                    {"id": row["id"], "file_path": normalized},
+                )
+                rows_changed += 1
+
+        if "photo" in table_names:
+            rows = sync_connection.execute(text('select id, path from "photo" where path is not null')).mappings().all()
+            for row in rows:
+                original = str(row["path"])
+                normalized = _canonical_relative_path(original, preferred_prefix="photos/")
+                if normalized is None or normalized == original:
+                    continue
+                sync_connection.execute(
+                    text('update "photo" set path = :path where id = :id'),
+                    {"id": row["id"], "path": normalized},
+                )
+                rows_changed += 1
+        return rows_changed
+
+    async with active_engine.begin() as connection:
+        rows_changed = await connection.run_sync(_reconcile)
+    if rows_changed:
+        logger.warning("Normalized %s media-path row(s) to canonical relative format", rows_changed)
+    return rows_changed
+
+
+async def reconcile_person_name_columns(*, engine: AsyncEngine | None = None) -> int:
+    """Backfill V5.1 Person name columns on existing databases."""
+    active_engine = engine or resolve_engine()
+    if not hasattr(active_engine, "begin"):
+        return 0
+
+    def _reconcile(sync_connection) -> int:
+        rows_changed = 0
+        inspector = sqlalchemy_inspect(sync_connection)
+        table_names = set(inspector.get_table_names())
+        if "person" not in table_names:
+            return 0
+        present_columns = {column["name"] for column in inspector.get_columns("person")}
+        if "last_name" not in present_columns:
+            sync_connection.execute(text('alter table "person" add column "last_name" varchar'))
+        if "given_names" not in present_columns:
+            sync_connection.execute(text('alter table "person" add column "given_names" varchar'))
+
+        query = (
+            text('select id, full_name, given_names, last_name from "person"')
+            if "full_name" in present_columns
+            else text('select id, null as full_name, given_names, last_name from "person"')
+        )
+        rows = sync_connection.execute(query).mappings().all()
+        for row in rows:
+            given_names = (str(row.get("given_names") or "")).strip()
+            last_name = (str(row.get("last_name") or "")).strip()
+            if given_names and last_name:
+                continue
+            tokens = [token for token in str(row.get("full_name") or "").split() if token]
+            if len(tokens) >= 2:
+                given_names, last_name = (" ".join(tokens[:-1]), tokens[-1])
+            elif len(tokens) == 1:
+                given_names = tokens[0]
+                last_name = tokens[0]
+            else:
+                given_names = "Unknown"
+                last_name = "Unknown"
+            sync_connection.execute(
+                text('update "person" set given_names = :given_names, last_name = :last_name where id = :id'),
+                {
+                    "id": row["id"],
+                    "given_names": given_names,
+                    "last_name": last_name,
+                },
+            )
+            rows_changed += 1
+        return rows_changed
+
+    async with active_engine.begin() as connection:
+        rows_changed = await connection.run_sync(_reconcile)
+    if rows_changed:
+        logger.warning("Backfilled V5.1 name columns for %s person row(s)", rows_changed)
+    return rows_changed
+
+
+def _canonical_relative_path(value: str, *, preferred_prefix: str) -> str | None:
+    normalized = value.strip().replace("\\", "/")
+    if not normalized:
+        return None
+
+    lowered = normalized.casefold()
+    if lowered.startswith(("http://", "https://", "data:")):
+        return None
+
+    if lowered.startswith("/uploads/"):
+        normalized = normalized[len("/uploads/") :]
+        lowered = normalized.casefold()
+    elif lowered.startswith("uploads/"):
+        normalized = normalized[len("uploads/") :]
+        lowered = normalized.casefold()
+    elif lowered.startswith("data/"):
+        normalized = normalized[len("data/") :]
+        lowered = normalized.casefold()
+
+    for prefix in ("documents/", "photos/", "persons/", "portraits/"):
+        marker = f"/{prefix}"
+        index = lowered.find(marker)
+        if index >= 0:
+            normalized = normalized[index + 1 :]
+            lowered = normalized.casefold()
             break
-            if not has_unique_source:
-                connection.execute(
-                    text(
-                        "CREATE UNIQUE INDEX IF NOT EXISTS "
-                        "ux_revision_source_id ON revision(source_id)"
+        if lowered.startswith(prefix):
+            break
+
+    if preferred_prefix == "persons/" and lowered.startswith("portraits/"):
+        normalized = "persons/" + normalized[len("portraits/") :]
+        lowered = normalized.casefold()
+
+    if not lowered.startswith(preferred_prefix):
+        return None
+
+    # Collapse any accidental "." segments while preserving relative semantics.
+    collapsed = Path(normalized).as_posix()
+    if collapsed.startswith("../") or collapsed == "..":
+        return None
+    return collapsed
+
+
+async def seed_registry_defaults(*, engine: AsyncEngine | None = None) -> None:
+    """Seed default registry rows for role and document type taxonomies."""
+    active_engine = engine or resolve_engine()
+    session_factory = async_sessionmaker(active_engine, class_=AsyncSession, expire_on_commit=False)
+
+    async with session_factory() as session:
+        role_keys = set((await session.exec(select(PersonRole.semantic_key))).all())
+        for semantic_key, label in BUILT_IN_PERSON_ROLES:
+            if semantic_key not in role_keys:
+                session.add(
+                    PersonRole(
+                        semantic_key=semantic_key,
+                        label=label,
+                        normalized_label=label.casefold(),
                     )
                 )
-                logger.warning(
-                    "Applied SQLite compatibility schema patch "
-                    "table=revision unique_index=ux_revision_source_id"
+
+        type_keys = set((await session.exec(select(DocumentType.semantic_key))).all())
+        for semantic_key, label in BUILT_IN_DOCUMENT_TYPES:
+            if semantic_key not in type_keys:
+                session.add(
+                    DocumentType(
+                        semantic_key=semantic_key,
+                        label=label,
+                        normalized_label=label.casefold(),
                     )
+                )
+
+        await session.commit()

src/transcription/db/registries.py

@@ -0,0 +1,20 @@
+"""Application-defined semantic registry entries."""
+
+from __future__ import annotations
+
+BUILT_IN_DOCUMENT_TYPES: tuple[tuple[str, str], ...] = (
+    ("book", "Book"),
+    ("letter", "Letter"),
+    ("postcard", "Postcard"),
+    ("photo", "Photo"),
+    ("journal", "Journal"),
+    ("form", "Form"),
+)
+
+BUILT_IN_PERSON_ROLES: tuple[tuple[str, str], ...] = (
+    ("author", "Author"),
+    ("recipient", "Recipient"),
+    ("mentioned", "Mentioned"),
+)
+
+AUTHOR_ROLE_SEMANTIC_KEY = "author"

src/transcription/db/runtime.py

@@ -1,18 +1,15 @@
 import logging
-from collections.abc import AsyncGenerator
-from contextlib import asynccontextmanager
-from contextvars import ContextVar
 from dataclasses import dataclass
-from functools import partial
 
 from sqlalchemy.ext.asyncio import AsyncEngine
 from sqlalchemy.ext.asyncio import async_sessionmaker
-from sqlalchemy.ext.asyncio import create_async_engine
 from sqlmodel.ext.asyncio.session import AsyncSession
-from sqlmodel.pool import StaticPool
 
 from ..config import Settings
 from ..config import get_settings
+from .engine import get_database_url
+from .engine import get_engine
+from .session import get_session_factory
 
 logger = logging.getLogger(__name__)
 
@@ -25,79 +22,41 @@ class DatabaseRuntime:
     session_factory: async_sessionmaker[AsyncSession]
 
 
-_runtime: ContextVar[DatabaseRuntime | None] = ContextVar("database_runtime", default=None)
+_runtime: DatabaseRuntime | None = None
+
+
+def get_database_runtime() -> DatabaseRuntime | None:
+    """Return the process-owned database runtime."""
+    return _runtime
 
 
 async def dispose_database_runtime() -> None:
     """Dispose lifespan-owned async database resources."""
-    runtime = _runtime.get()
+    global _runtime
+    runtime = _runtime
     if runtime is None:
         return
     await runtime.engine.dispose()
-    _runtime.set(None)
-
-
-def _to_async_database_url(database_url: str) -> str:
-    """Normalize configured database URL to an async SQLAlchemy driver URL."""
-    if database_url.startswith("sqlite://") and not database_url.startswith("sqlite+aiosqlite://"):
-        return database_url.replace("sqlite://", "sqlite+aiosqlite://", 1)
-    if database_url.startswith("postgresql://") and not database_url.startswith("postgresql+asyncpg://"):
-        return database_url.replace("postgresql://", "postgresql+asyncpg://", 1)
-    return database_url
-
-
-def _build_engine(settings: Settings) -> AsyncEngine:
-    database_url = _to_async_database_url(settings.database_url)
-    engine_factory = partial(
-        create_async_engine,
-        url=database_url,
-        echo=False,
-        pool_pre_ping=True,
-    )
-
-    if database_url.startswith("sqlite"):
-        sqlite_connect_settings = {"check_same_thread": settings.sqlite_check_same_thread}
-        engine_factory = partial(engine_factory, connect_args=sqlite_connect_settings)
-        if ":memory:" in database_url:
-            engine_factory = partial(engine_factory, poolclass=StaticPool)
-
-    return engine_factory()
+    _runtime = None
 
 
 def initialize_database_runtime(*, settings: Settings | None = None) -> DatabaseRuntime:
     """Initialize lifespan-owned async DB resources once per process."""
-    runtime = _runtime.get()
+    global _runtime
+    active_settings = settings or get_settings()
+    database_url = get_database_url(active_settings)
+    runtime = _runtime
     if runtime is not None:
+        runtime_url = runtime.engine.url.render_as_string(hide_password=False)
+        if runtime_url != database_url:
+            raise RuntimeError(
+                f"Database runtime is already initialized for a different database: {runtime_url!r} != {database_url!r}"
+            )
         return runtime
 
-    active_settings = settings or get_settings()
-    engine = _build_engine(active_settings)
-    session_factory = async_sessionmaker(engine, class_=AsyncSession, expire_on_commit=False)
+    engine = get_engine(database_url)
+    session_factory = get_session_factory(database_url)
     runtime = DatabaseRuntime(engine=engine, session_factory=session_factory)
-    _runtime.set(runtime)
+    _runtime = runtime
     logger.debug("Initialized async database runtime for database_url=%s", engine.url)
     return runtime
-
-
-def get_engine(settings: Settings | None = None) -> AsyncEngine:
-    """Return the current async SQLAlchemy engine."""
-    runtime = _runtime.get() or initialize_database_runtime(settings=settings)
-    return runtime.engine
-
-
-def get_session_factory(settings: Settings | None = None) -> async_sessionmaker[AsyncSession]:
-    """Return the shared async session factory."""
-    runtime = _runtime.get() or initialize_database_runtime(settings=settings)
-    return runtime.session_factory
-
-
-@asynccontextmanager
-async def get_session(
-    *,
-    settings: Settings | None = None,
-    session_factory: async_sessionmaker[AsyncSession] | None = None,
-) -> AsyncGenerator[AsyncSession]:
-    """Yield a database session and ensure cleanup."""
-    active_session_factory = session_factory or get_session_factory(settings)
-    async with active_session_factory() as session:
-        yield session

src/transcription/db/session.py

@@ -0,0 +1,104 @@
+from collections.abc import AsyncGenerator
+from contextlib import asynccontextmanager
+from typing import Annotated
+
+from fastapi import Depends
+from sqlalchemy.ext.asyncio import async_sessionmaker
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from ..config import Settings
+from ..config import get_settings
+from .engine import dispose_engine
+from .engine import get_database_url
+from .engine import get_engine
+
+type SessionFactory = async_sessionmaker[AsyncSession]
+
+
+_SESSION_FACTORIES: dict[str, SessionFactory] = {}
+
+
+def get_session_factory(database_url: str) -> SessionFactory:
+    """Return the process-wide session factory for ``database_url``."""
+    factory = _SESSION_FACTORIES.get(database_url)
+    if factory is None:
+        factory = async_sessionmaker(
+            bind=get_engine(database_url),
+            class_=AsyncSession,
+            expire_on_commit=False,
+        )
+        _SESSION_FACTORIES[database_url] = factory
+    return factory
+
+
+def resolve_session_factory(
+    database_url: str | None = None,
+    *,
+    settings: Settings | None = None,
+) -> SessionFactory:
+    if database_url is not None:
+        return get_session_factory(database_url)
+    if settings is None:
+        from .runtime import get_database_runtime
+
+        runtime = get_database_runtime()
+        if runtime is not None:
+            return runtime.session_factory
+    return get_session_factory(get_database_url(settings or get_settings()))
+
+
+type SessionFactoryDep = Annotated[SessionFactory, Depends(resolve_session_factory)]
+
+
+async def dispose_session_factory(database_url: str) -> None:
+    """Drop the session factory and engine for ``database_url`` only."""
+    _SESSION_FACTORIES.pop(database_url, None)
+    await dispose_engine(database_url)
+
+
+@asynccontextmanager
+async def session_scope(
+    *,
+    settings: Settings | None = None,
+    database_url: str | None = None,
+    session_factory: SessionFactory | None = None,
+    session: AsyncSession | None = None,
+) -> AsyncGenerator[AsyncSession]:
+    if session is not None:
+        yield session
+        return
+
+    active_session_factory = session_factory or resolve_session_factory(
+        database_url,
+        settings=settings,
+    )
+    async with active_session_factory() as owned_session:
+        yield owned_session
+
+
+type SessionScopeDep = Annotated[AsyncSession, Depends(session_scope)]
+
+
+@asynccontextmanager
+async def transaction_scope(
+    *,
+    settings: Settings | None = None,
+    database_url: str | None = None,
+    session_factory: SessionFactory | None = None,
+    session: AsyncSession | None = None,
+) -> AsyncGenerator[AsyncSession]:
+    if session is not None:
+        if not session.in_transaction():
+            raise RuntimeError("A supplied session must have an active transaction")
+        yield session
+        return
+
+    active_session_factory = session_factory or resolve_session_factory(
+        database_url,
+        settings=settings,
+    )
+    async with active_session_factory.begin() as owned_session:
+        yield owned_session
+
+
+type TransactionScopeDep = Annotated[AsyncSession, Depends(transaction_scope)]

src/transcription/errors.py

@@ -2,12 +2,15 @@
 
 from __future__ import annotations
 
+import logging
 from dataclasses import dataclass
 from datetime import UTC
 from datetime import datetime
 from enum import StrEnum
 from uuid import uuid4
 
+logger = logging.getLogger(__name__)
+
 
 class ErrorCategory(StrEnum):
     """Stable error categories defined by docs/error_handling.md."""
@@ -17,6 +20,7 @@ class ErrorCategory(StrEnum):
     NOT_FOUND = "not_found_error"
     CONFLICT = "conflict_error"
     EXTERNAL_PROVIDER = "external_provider_error"
+    EXTERNAL_TIMEOUT = "external_timeout_error"
     PROCESSING = "processing_error"
     INFRA_TRANSIENT = "infrastructure_transient_error"
     INFRA_PERSISTENT = "infrastructure_persistent_error"
@@ -28,6 +32,11 @@ def new_error_id() -> str:
     return uuid4().hex[:8]
 
 
+def exception_detail(exc: BaseException) -> str:
+    """Return internal-only root-cause text for persisted diagnostics."""
+    return f"{type(exc).__name__}: {exc}"
+
+
 class AppError(RuntimeError):
     """Base application error carrying user-safe handling metadata."""
 
@@ -39,6 +48,7 @@ class AppError(RuntimeError):
         suggestion: str = "Retry once. If it persists, review logs and report the error reference id.",
         retriable: bool = False,
         error_id: str | None = None,
+        detail: str | None = None,
     ) -> None:
         super().__init__(message)
         self.message = message
@@ -46,6 +56,10 @@ class AppError(RuntimeError):
         self.suggestion = suggestion
         self.retriable = retriable
         self.error_id = error_id or new_error_id()
+        # Internal-only diagnostic text. Persisted to evidence and logs, never rendered
+        # to users or serialized into API envelopes, because it may embed local
+        # filesystem paths and other infrastructure detail.
+        self.detail = detail
 
 
 @dataclass(frozen=True)
@@ -59,11 +73,28 @@ class ErrorEnvelope:
     timestamp: str
 
 
+def canonical_error_category(error: AppError) -> str:
+    """Map internal categories to canonical API/UI envelope categories."""
+    mapping: dict[ErrorCategory, str] = {
+        ErrorCategory.VALIDATION: "validation",
+        ErrorCategory.USER_INPUT: "validation",
+        ErrorCategory.NOT_FOUND: "not_found",
+        ErrorCategory.CONFLICT: "conflict",
+        ErrorCategory.EXTERNAL_PROVIDER: "external",
+        ErrorCategory.EXTERNAL_TIMEOUT: "timeout",
+        ErrorCategory.INFRA_TRANSIENT: "timeout",
+        ErrorCategory.PROCESSING: "internal",
+        ErrorCategory.INFRA_PERSISTENT: "internal",
+        ErrorCategory.INTERNAL_UNEXPECTED: "internal",
+    }
+    return mapping.get(error.category, "internal")
+
+
 def build_error_envelope(error: AppError) -> ErrorEnvelope:
     """Build an API-safe response envelope from an AppError."""
     return ErrorEnvelope(
         error_id=error.error_id,
-        category=error.category.value,
+        category=canonical_error_category(error),
         message=error.message,
         suggestion=error.suggestion,
         timestamp=datetime.now(UTC).isoformat(),
@@ -71,15 +102,43 @@ def build_error_envelope(error: AppError) -> ErrorEnvelope:
 
 
 def classify_unexpected_error(exc: Exception, *, operation: str) -> AppError:
-    """Normalize unknown exceptions into internal_unexpected_error."""
-    return AppError(
-        f"Unexpected error during {operation}: {exc}",
+    """Normalize unknown exceptions into internal_unexpected_error.
+
+    The exception text is deliberately excluded from ``message``. ``AppError.message``
+    is rendered directly to users by the UI error presenter and is serialized into API
+    responses by :func:`build_error_envelope`, and unexpected exceptions routinely embed
+    local filesystem paths (SQLAlchemy ``OperationalError`` carries the database path,
+    ``OSError`` carries the storage root). Leaking those is forbidden by
+    ``.github/instructions/error-handling.instructions.md``.
+
+    The detail is preserved on ``AppError.detail`` and logged against ``error_id``. That
+    keeps the root cause in evidence records and operator logs, which are internal, while
+    keeping it out of user-facing and API-facing text.
+    """
+    error = AppError(
+        f"Unexpected error during {operation}.",
         category=ErrorCategory.INTERNAL_UNEXPECTED,
         suggestion="Retry once. If it persists, review logs and report the error reference id.",
         retriable=False,
+        detail=exception_detail(exc),
     )
+    logger.error(
+        "Unexpected error operation=%s error_id=%s",
+        operation,
+        error.error_id,
+        exc_info=exc,
+    )
+    return error
 
 
 def format_error_detail(error: AppError) -> str:
-    """Return a compact persisted failure string for transcript.error_detail."""
-    return f"[{error.category.value}] {error.message} | suggestion={error.suggestion} | error_id={error.error_id}"
+    """Return a compact persisted failure string for transcript.error_detail.
+
+    This is internal provenance, not user-facing output, so it carries
+    ``AppError.detail`` (the root cause) in addition to the user-safe message.
+    """
+    parts = [f"[{error.category.value}] {error.message}"]
+    if error.detail:
+        parts.append(f"detail={error.detail}")
+    parts.extend((f"suggestion={error.suggestion}", f"error_id={error.error_id}"))
+    return " | ".join(parts)

src/transcription/models.py

@@ -1,110 +0,0 @@
-"""SQLModel domain models for the transcription system.
-
-Core V1 lifecycle:
-    Document  ->  one-to-many  ->  Source
-    Document  ->  one-to-many  ->  Job
-    Source    ->  one-to-one?  ->  Revision (optional)
-"""
-
-from datetime import UTC
-from datetime import datetime
-from enum import StrEnum
-from typing import Optional
-from uuid import UUID
-from uuid import uuid4
-
-from sqlalchemy import UniqueConstraint
-from sqlmodel import Field
-from sqlmodel import Relationship
-from sqlmodel import SQLModel
-
-
-class JobStatus(StrEnum):
-    QUEUED = "queued"
-    PROCESSING = "processing"
-    TRANSCRIBED = "transcribed"
-    FAILED = "failed"
-
-
-class Document(SQLModel, table=True):
-    """An historical document."""
-
-    id: UUID = Field(default_factory=uuid4, primary_key=True)
-    name: str
-
-    # Relationships
-    jobs: list["Job"] = Relationship(back_populates="document")
-    sources: list["Source"] = Relationship(back_populates="document")
-
-
-class Source(SQLModel, table=True):
-    """A document source (image or PDF)."""
-
-    id: UUID = Field(default_factory=uuid4, primary_key=True)
-    document_id: UUID = Field(foreign_key="document.id")
-    job_id: UUID = Field(foreign_key="job.id")
-    upload_name: str
-    """The filename of the source that was uploaded for transcription."""
-    filename: str
-    """The system generated unique source name."""
-    file_path: str
-    """The location where the sources are stored on the local filesystem."""
-    date_uploaded: datetime = Field(default_factory=lambda: datetime.now(UTC))
-
-    # Relationships
-    document: Optional["Document"] = Relationship(back_populates="sources")
-    job: Optional["Job"] = Relationship(back_populates="sources")
-    revision: Optional["Revision"] = Relationship(
-        back_populates="source",
-        sa_relationship_kwargs={"uselist": False},
-    )
-
-
-class Job(SQLModel, table=True):
-    """A transcription job tied to a single document."""
-
-    id: UUID = Field(default_factory=uuid4, primary_key=True)
-    document_id: UUID = Field(foreign_key="document.id")
-    status: JobStatus = Field(default=JobStatus.QUEUED)
-    retry_count: int = Field(default=0, ge=0)
-    date_created: datetime = Field(default_factory=lambda: datetime.now(UTC))
-    date_updated: datetime = Field(default_factory=lambda: datetime.now(UTC))
-    provider: str | None = None
-    """Name of the transcription provider used to generate this transcript."""
-    model: str | None = None
-    """Model identifier used to generate this transcript."""
-    prompt_name: str | None = None
-    """Name of the prompt used to generate this transcript."""
-    text: str | None = None
-    """The transcribed text. This may be None if the job failed or is still in progress."""
-    error_detail: str | None = None
-    """Details of any error that occurred during transcription."""
-
-    # Relationships
-    document: Optional["Document"] = Relationship(back_populates="jobs")
-    sources: list["Source"] = Relationship(back_populates="job")
-
-    @property
-    def filename(self) -> str:
-        """Return the filename of the associated source, when available."""
-        if not self.sources:
-            return "unknown"
-        return self.sources[0].filename
-
-
-class Revision(SQLModel, table=True):
-    """A revision of a transcription text."""
-
-    id: UUID = Field(default_factory=uuid4, primary_key=True)
-    source_id: UUID = Field(foreign_key="source.id")
-    """ID for the associated source."""
-    revision: int = Field(default=1, ge=1)
-    """Revision number of this transcription revision, starting at 1."""
-    text: str
-    """The revised text."""
-    date_created: datetime = Field(default_factory=lambda: datetime.now(UTC))
-
-    __table_args__ = (UniqueConstraint("source_id", name="uq_revision_source_id"),)
-
-    # Relationships
-    source: Optional["Source"] = Relationship(back_populates="revision")

src/transcription/providers/__init__.py

@@ -4,10 +4,15 @@ from transcription.config import Provider
 from transcription.config import Settings
 from transcription.config import get_settings
 from transcription.providers.base import ProviderAuthError
+from transcription.providers.base import ProviderCallEvidence
 from transcription.providers.base import ProviderError
 from transcription.providers.base import ProviderResponseError
+from transcription.providers.base import TranscriptionMetadata
 from transcription.providers.base import TranscriptionProvider
 from transcription.providers.base import TranscriptionResult
+from transcription.providers.evidence import RequestManifest
+from transcription.providers.evidence import SourceEvidenceReference
+from transcription.providers.evidence import TransportEvidence
 from transcription.providers.openrouter import OpenRouterTranscriptionProvider
 
 
@@ -23,9 +28,14 @@ def get_transcription_provider(*, settings: Settings | None = None) -> Transcrip
 __all__ = [
     "OpenRouterTranscriptionProvider",
     "ProviderAuthError",
+    "ProviderCallEvidence",
     "ProviderError",
     "ProviderResponseError",
+    "RequestManifest",
+    "SourceEvidenceReference",
+    "TranscriptionMetadata",
     "TranscriptionProvider",
     "TranscriptionResult",
+    "TransportEvidence",
     "get_transcription_provider",
 ]