最近的服务
处理进展

影响范围

release observability from the V2.3 audit, limited to surfacing async task trace continuity in admin monitoring.

这轮变更

extended admin monitoring with `api_worker_trace_correlation_24h` and `recent_traced_tasks`, added the response schema fields, and expanded the worker trace regression test to verify admin monitoring sees the same API-to-worker trace id chain.

验证结果

`python -m pytest apps/api/tests/test_worker_request_trace.py -q` passed with 1 test, confirming admin monitoring now exposes trace-correlated tasks and worker runs.

影响范围

release observability from the V2.3 audit, limited to extending trace correlation from the API request layer into async task creation and worker execution logs.

这轮变更

added `tasks.trace_id` plus a migration, persisted the incoming request `trace_id` when creating tasks, exposed `trace_id` on task reads, and updated the worker scheduler to reuse that trace context so `worker_task_started`, dispatch, and completion logs carry the same trace id as the originating API request.

验证结果

`python -m pytest apps/api/tests/test_observability.py apps/api/tests/test_worker_request_trace.py -q` passed with 4 tests, confirming API trace headers still work and worker lifecycle logs now reuse the original task trace id.

影响范围

release observability from the V2.3 audit, limited to server-side trace correlation, persisted API request telemetry, and a 24-hour admin summary of request health.

这轮变更

added the `api_request_telemetry` model plus migration, extended API observability context to carry `trace_id`, updated the request middleware to emit `X-Trace-ID`, persist per-request telemetry, and log trace-aware completion and failure events, and extended admin monitoring with `api_request_telemetry_24h_summary`, `api_request_telemetry_24h_by_status`, and `api_trace_correlation_24h`.

验证结果

`python -m pytest apps/api/tests/test_observability.py -q` passed with 3 tests, confirming trace headers, persisted request telemetry, exception logging, and admin monitoring aggregation.

影响范围

release observability from the V2.3 audit, limited to shipping a real browser-side Core Web Vitals reporter, persisting samples in the API, and surfacing a 24-hour admin summary.

这轮变更

added the `web_vital_samples` model plus migration, added `/api/observability/web-vitals` for persisted ingest, extended admin monitoring with `web_vitals_24h_summary` and `web_vitals_24h_by_metric`, added a `WebVitalsReporter` client component to the web app root layout using Next web-vitals hooks, and added targeted API regression coverage for ingest plus monitoring output.

验证结果

`python -m pytest apps/api/tests/test_web_vitals_rum.py -q` passed with 1 test, and `npm run build` passed in `apps/web`, confirming the ingest path and browser reporter compile cleanly.

影响范围

release-platform governance from the V2.3 audit, limited to centralizing a controlled set of admin release flags and wiring one existing admin surface to respect them.

这轮变更

added the `feature_flag_entries` model plus migration, added `apps/api/app/services/feature_flags.py` with typed flag definitions and runtime lookups, added `/api/admin/feature-flags` read/write routes, wired `/api/admin/monitoring/overview` to the `release_admin_monitoring` flag, and audited flag mutations through the existing admin audit log.

验证结果

`python -m pytest apps/api/tests/test_feature_flags.py -q` passed with 1 test, confirming feature flag updates are persisted, admin-visible, audited, and enforced by the admin monitoring route.

影响范围

release-platform governance from the V2.3 audit, limited to centralizing a controlled set of runtime settings and wiring the existing retention policy path to use them.

这轮变更

added the `system_config_entries` model plus migration, added `apps/api/app/services/system_config.py` with typed config definitions and retention policy lookups, added `/api/admin/system-config` read/write routes, wired monitoring and retention cleanup to use persisted retention overrides, and audited config mutations through the existing admin audit log.

验证结果

`python -m pytest apps/api/tests/test_system_config_center.py -q` passed with 1 test, confirming config updates are persisted, admin-visible, audited, and applied by retention monitoring and cleanup.

影响范围

admin governance from the V2.3 audit, limited to recording key mutating admin actions and exposing a read path for audit review.

这轮变更

added the `admin_operation_audits` model plus migration, wired persisted audit records into admin task retry, task cancel, password recovery status update, and run-once execution routes, and added `/api/admin/audit-log` for review.

验证结果

`python -m pytest apps/api/tests/test_admin_operation_audit.py -q` passed with 1 test, confirming the audited admin actions are written to storage and readable through the new audit endpoint.

影响范围

operational recovery from the V2.3 audit, limited to surfacing the latest failed backup and restore evidence together with any scheduled plans that failed recently inside the existing admin monitoring endpoint.

这轮变更

extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_failure_health` with latest failed operation timestamps, 7-day failure counts, and recent failed scheduled plan names, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the new failure-health summary against a controlled backup history file.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now exposes backup and restore failure health alongside the existing recovery summaries.

影响范围

operational recovery from the V2.3 audit, limited to surfacing the latest failed backup and restore evidence and recent failure counts inside the existing admin monitoring endpoint.

这轮变更

extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_failure_health` with latest failed operation timestamps, 7-day failure counts, and recent failed scheduled plan names, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the new failure-health summary against a controlled backup history file.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now exposes backup and restore failure health alongside the existing recovery summaries.

影响范围

operational recovery from the V2.3 audit, limited to recording failed backup and restore attempts and surfacing a 30-day failure trend inside the existing monitoring and drill path.

这轮变更

extended `infra/deploy/backup_restore_sqlite.py` to write failed `backup`, `restore`, and `scheduled_backup` attempts to history, extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_failures_30d_daily`, and extended backup and admin monitoring tests to verify both the failure history record and the new 30-day failure trend.

验证结果

`python -m pytest apps/api/tests/test_backup_restore_snapshot.py apps/api/tests/test_auth_wallet_tasks.py -q -k "restore_failure_is_written_to_history or backup_recovery_evidence"` passed with 2 tests, confirming failed restore attempts are persisted to history and surfaced in admin monitoring.

影响范围

operational recovery from the V2.3 audit, limited to surfacing a longer-window daily restore-readiness view inside the existing admin monitoring endpoint.

这轮变更

extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_readiness_30d_daily` with per-day backup evidence counts, restore counts, scheduled plan coverage counts, and `recovery_ready` flags, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the new 30-day readiness trend against a controlled backup history file.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now exposes the 30-day restore-readiness trend.

影响范围

operational recovery from the V2.3 audit, limited to surfacing a 7-day daily history of backup, scheduled-backup, and restore evidence inside the existing admin monitoring endpoint.

这轮变更

extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_7d_daily` with per-day backup, scheduled-backup, and restore counts, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the daily history against a controlled backup history file.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now exposes the 7-day backup and restore history trend.

影响范围

operational recovery from the V2.3 audit, limited to reporting whether recent backup evidence, scheduled plan coverage, and restore drill freshness are sufficient inside the existing admin monitoring endpoint.

这轮变更

extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_readiness` with backup freshness window checks, restore-readiness window checks, schedule coverage counts, and an overall `recovery_ready` flag, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the new readiness summary against a controlled backup history file.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now surfaces restore readiness alongside backup freshness and schedule coverage.

影响范围

operational recovery from the V2.3 audit, limited to summarizing backup freshness, restore freshness, and missing scheduled runs inside the existing admin monitoring endpoint.

这轮变更

extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `backup_recovery_health` with latest backup/scheduled-backup/restore timestamps, 7-day counts, and `plans_missing_recent_run`, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the new health summary against a controlled backup history file.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now summarizes backup and restore freshness plus schedule coverage.

影响范围

operational recovery from the V2.3 audit, limited to exposing the repo backup schedule and recent recovery history through the existing admin monitoring endpoint.

这轮变更

added `backup_history_file` and `backup_plan_file` settings in `apps/api/app/config.py`, extended `apps/api/app/services/tasks.py` and `apps/api/app/schemas/task.py` so `/api/admin/monitoring/overview` now returns `scheduled_backup_plans` and `recent_backup_recovery_runs`, and extended `apps/api/tests/test_auth_wallet_tasks.py` to verify the endpoint reads a named plan plus recent backup, scheduled-backup, and restore history records.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q -k backup_recovery_evidence` passed with 1 test, confirming admin monitoring now surfaces the configured backup plan and recent recovery evidence.

影响范围

operational recovery from the V2.3 audit, limited to giving the SQLite drill a repo-backed named schedule and archive retention policy.

这轮变更

extended `infra/deploy/backup_restore_sqlite.py` with named scheduled plan loading, a `scheduled-backup` command, and retention pruning, added `infra/deploy/backup_plan.example.json` as the repo baseline plan, updated `infra/deploy/README.md` and `infra/deploy/LAUNCH_CHECKLIST.md` to document scheduled execution, and extended `apps/api/tests/test_backup_restore_snapshot.py` to verify repeated scheduled runs keep the expected number of archives and emit schedule history records.

验证结果

`python -m pytest apps/api/tests/test_backup_restore_snapshot.py -q` passed with 2 tests, confirming the SQLite drill now supports named scheduled backups with retention pruning and history evidence.

影响范围

operational recovery from the V2.3 audit, limited to turning the SQLite backup drill into a repo-backed evidence trail for operators.

这轮变更

extended `infra/deploy/backup_restore_sqlite.py` so backup and restore append JSONL records and expose a `history` command, updated `infra/deploy/README.md` and `infra/deploy/LAUNCH_CHECKLIST.md` to point operators at `backup_restore_history.jsonl`, and extended `apps/api/tests/test_backup_restore_snapshot.py` to verify both backup and restore leave inspectable history records.

验证结果

`python -m pytest apps/api/tests/test_backup_restore_snapshot.py -q` passed with 1 test, confirming the SQLite drill now leaves a reusable JSONL history trail for both backup and restore operations.

影响范围

operational recovery from the V2.3 audit, limited to the repo default SQLite deployment path and its local artifact directories.

这轮变更

added `infra/deploy/backup_restore_sqlite.py` with `backup` and `restore` commands that snapshot `data/taskgpu.db` plus `delivery_artifacts/` and `imported_files/` into a zip archive, documented the drill in `infra/deploy/README.md`, and added `apps/api/tests/test_backup_restore_snapshot.py` to prove a backup can be restored into a fresh directory with database rows and artifact files intact.

验证结果

`python -m pytest apps/api/tests/test_backup_restore_snapshot.py -q` passed with 1 test, confirming the SQLite snapshot archive can be restored with both database content and artifact payloads preserved.

影响范围

backend platform maturity from the V2.3 audit, limited to replacing raw schema bootstrap with a tracked migration path and one baseline version.

这轮变更

added `apps/api/app/migration_runner.py` to discover and apply versioned migrations into a tracked `schema_migrations` table, added `apps/api/migrations/versions/0001_baseline.py` plus package scaffolding, switched `apps/api/app/database.py` so `init_db()` now runs migrations instead of `Base.metadata.create_all()`, added `tools/database/run_api_migrations.py` as a CLI entrypoint, and added `apps/api/tests/test_migration_runner.py` to lock the baseline migration behavior on a fresh SQLite database.

验证结果

`python -m pytest apps/api/tests/test_migration_runner.py -q` passed with 1 test, confirming a fresh database is created through the migration runner and records the applied baseline migration exactly once.

影响范围

release-platform security from the V2.3 audit, limited to adding one repo-local dependency hygiene gate and enforcing it in GitHub Actions.

这轮变更

added `tools/security/run_dependency_audit.py` to validate bounded Python requirements, reject direct or editable requirement sources, require a modern Node lockfile, and enforce https plus integrity metadata for resolved npm packages, then wired the runner into the existing `security-baseline` job in `.github/workflows/ci.yml`.

验证结果

`python tools/security/run_dependency_audit.py` passed and reported zero violations across `apps/api/requirements.txt` and `apps/web/package-lock.json`.

影响范围

security baseline from the V2.3 audit, limited to adding a narrow HTTP-level dynamic security check and enforcing it in GitHub Actions.

这轮变更

added `tools/security/run_repo_dast.py` to exercise `/api/health` and `/api/v1/health` through FastAPI `TestClient` and assert `X-Request-ID`, `X-Content-Type-Options`, `X-Frame-Options`, `Referrer-Policy`, and `Cache-Control`, hardened `apps/api/app/main.py` to set those headers on normal and exception responses, and wired the runner into `.github/workflows/ci.yml` before the API and worker tests.

验证结果

`python tools/security/run_repo_dast.py` passed and reported both health endpoints as passing the repo DAST baseline.

影响范围

security baseline from the V2.3 audit, limited to adding a narrow static-analysis pass and enforcing it in GitHub Actions.

这轮变更

added `tools/security/run_repo_sast.py` to scan repo source for unapproved `shell=True`, `verify=False`, and dynamic `eval`/`exec` usage with a documented allowlist for the existing code editor executor, and added a new `security-baseline` job in `.github/workflows/ci.yml` to run the scanner on every push and pull request.

验证结果

`python tools/security/run_repo_sast.py` passed and reported zero violations with one documented allowlisted finding.

影响范围

account security from the V2.3 audit, limited to detecting unusual login fingerprints and exposing the resulting alerts to signed-in users.

这轮变更

added `AuthSecurityAlert` persistence in `apps/api/app/models.py`, exposed `GET /api/auth/security-alerts` plus alert serialization in `apps/api/app/routes/auth.py` and `apps/api/app/schemas/auth.py`, and updated the login flow to create `new_ip` and `new_device` alerts when a successful login arrives from an unseen fingerprint without flagging refresh-token rotation.

验证结果

`python -m pytest apps/api/tests/test_login_alerts.py apps/api/tests/test_session_management.py apps/api/tests/test_refresh_token.py -q` passed with 6 tests, confirming unusual logins now create user-visible alerts while the existing session and refresh flows still behave correctly.

影响范围

backend platform maturity from the V2.3 audit, limited to closing the missing API versioning gap without breaking the current unversioned `/api/*` clients.

这轮变更

added an API version alias middleware in `apps/api/app/main.py` that rewrites `/api/v1/*` requests onto the existing `/api/*` routes, and added `apps/api/tests/test_api_versioning.py` to lock health, register, and login access through the new versioned path.

验证结果

`python -m pytest apps/api/tests/test_api_versioning.py -q` passed with 2 tests, confirming the repo now serves a working `/api/v1` compatibility layer alongside the current endpoints.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation-hit history grouped by day, stage, and moderation rule through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate blocked planner/reviewer moderation events into `ai_moderation_hits_7d_daily_by_stage_flag`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the daily stage-rule history payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation stage+rule daily history without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day governance-enforcement history grouped by day, stage, and fallback error through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_7d_daily_by_stage_error`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the daily stage-error history payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI governance stage+error daily history without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day governance-enforcement totals grouped by stage and fallback error through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_7d_by_stage_error`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the stage-error summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI governance stage+error totals without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation outcomes grouped by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate planner/reviewer moderation events into `ai_moderation_outcomes_7d_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the 7-day moderation-outcome summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation outcome totals without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation-hit totals grouped by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate blocked planner/reviewer moderation events into `ai_moderation_hits_7d_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the 7-day moderation-stage summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation stage totals without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation-hit history grouped by day and moderation rule through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate blocked planner/reviewer moderation events into `ai_moderation_hits_7d_daily_by_flag`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the daily moderation-rule history payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation rule history without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation outcomes grouped by day and stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate planner/reviewer moderation events into `ai_moderation_outcomes_7d_daily_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the daily moderation-outcome payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation outcomes without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation-hit totals grouped by moderation rule through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate blocked planner/reviewer moderation events into `ai_moderation_hits_7d_by_flag`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the 7-day moderation-rule summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation rule totals without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day moderation-hit history grouped by day and stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate blocked planner/reviewer moderation events into `ai_moderation_hits_7d_daily_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the daily moderation-history payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day AI moderation hit history without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day governance-enforcement totals grouped by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_7d_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the stage-summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day governance block stage totals without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day governance-enforcement history grouped by day and fallback error through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_7d_daily_by_error`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the daily error-history payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day governance block daily error history without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing 7-day governance-enforcement counts grouped by fallback error through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_7d_by_error`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the error summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day governance block error summaries without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing daily governance-enforcement history by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_7d_daily_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed older governance block events and lock the daily history payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes 7-day governance block history without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing recent governance-enforcement trend changes by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to compare recent `governance_incomplete` fallback events between the current one-hour window and the earlier baseline window and expose `ai_governance_block_trend_deltas_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes governance block trend deltas without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing recent governance-enforcement fallback events by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `gpt_workflow` fallback events with `fallback_reason=governance_incomplete` into `ai_governance_blocks_24h_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed governance block events and lock the payload without relying on warning-stage ordering.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes recent AI governance blocks without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to blocking GPT planning or review when required eval or red-team coverage is missing for that stage.

这轮变更

extended `apps/api/app/services/gpt_workflow.py` with governance coverage checks for eval and red-team baselines before model execution, returning a `governance_incomplete` fallback when coverage is missing, and updated `apps/api/tests/test_gpt_workflow.py` with regression cases that verify the model is not called when governance coverage is absent.

验证结果

`python -m pytest apps/api/tests/test_gpt_workflow.py -q` passed with 8 tests, confirming planning and review now hard-block on missing AI governance coverage without regressing the existing GPT workflow paths.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing one admin-visible summary of prompt, moderation, eval, and red-team baseline coverage through the existing monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to expose `ai_governance_summary`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes the AI governance summary without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing red-team scenario and case coverage by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate `RED_TEAM_SCENARIO_REGISTRY` into `ai_red_team_coverage_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes AI red-team coverage summaries without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing the in-repo red-team scenario baseline through the existing admin monitoring overview.

这轮变更

added `RED_TEAM_SCENARIO_REGISTRY` in `apps/api/app/services/gpt_workflow.py`, extended `apps/api/app/services/tasks.py` to expose `ai_red_team_inventory`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes the AI red-team inventory without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing recent moderation blocked-rate deltas by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to compare recent planner and reviewer moderation outcomes in the last hour versus the prior baseline window and expose `ai_moderation_trend_deltas_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed baseline moderation events and lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes AI moderation trend deltas without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing eval focus coverage by focus term, case count, and covered stages through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate the in-repo eval registry into `ai_eval_focus_coverage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes AI eval focus summaries without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing eval-set and case-count coverage by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate the in-repo eval registry into `ai_eval_coverage_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes AI eval coverage summaries without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing recent moderation totals, blocked counts, passed counts, and blocked rates by stage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate recent `planner` and `reviewer` moderation outcomes into `ai_moderation_outcomes_24h_by_stage`, updated `apps/api/app/schemas/task.py` with the new monitoring field, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed passed moderation events and lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes AI moderation outcome rates without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing recent blocked moderation events by stage and flag through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/tasks.py` to aggregate blocked `planner` and `reviewer` task events into `ai_moderation_hits_24h_by_stage` and `ai_moderation_hits_24h_by_flag`, updated `apps/api/app/schemas/task.py` with the new monitoring fields, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed blocked moderation events and lock the new payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes AI moderation hit summaries without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to defining a baseline in-repo eval dataset registry and surfacing it through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/services/gpt_workflow.py` with a versioned `EVAL_DATASET_REGISTRY`, updated `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `ai_eval_inventory` with eval set ID, version, stage, case count, and focus areas, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new monitoring payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes the AI eval inventory without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing the local moderation ruleset and category coverage through the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `ai_moderation_inventory` with moderation category, ruleset version, and pattern count, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new monitoring payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes the AI moderation inventory without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to adding a repo-local moderation and output-safety filter around the existing GPT planning and review workflow.

这轮变更

extended `apps/api/app/services/gpt_workflow.py` so planning and review now apply versioned local moderation rules to both request inputs and model outputs, return `moderation_version`, `moderation_status`, and `moderation_flags`, and block flagged content with safe fallback results; updated `apps/api/tests/test_gpt_workflow.py` to lock the new moderation metadata and blocking behavior with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_gpt_workflow.py -q` passed with 6 tests, confirming the moderation hook blocks flagged inputs/outputs while preserving the existing prompt metadata contract.

影响范围

AI-governance remediation from the V2.3 audit, limited to surfacing the existing prompt registry and version metadata through the admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `ai_prompt_inventory` with `stage`, `prompt_id`, `prompt_version`, and configured model for the planning and review prompts, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new monitoring payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes the AI prompt inventory without regressing the existing admin rollups.

影响范围

AI-governance remediation from the V2.3 audit, limited to introducing a minimal prompt registry and version metadata for the existing planning and review prompts.

这轮变更

extended `apps/api/app/services/gpt_workflow.py` so planning and review results now carry `prompt_stage`, `prompt_id`, and `prompt_version` from a central prompt registry in both success and fallback paths, and added `apps/api/tests/test_gpt_workflow.py` to lock the new prompt metadata behavior with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_gpt_workflow.py -q` passed with 4 tests, confirming the GPT workflow now emits stable prompt version metadata for planning and review without changing the existing output contract.

影响范围

data-governance remediation from the V2.3 audit, limited to aggregating the current governance inventory into a compact coverage summary for admin review.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `data_governance_summary` with tracked-entity count, classification coverage, retention coverage, audit coverage, and entities without retention policy, and expanded `apps/api/tests/test_auth_wallet_tasks.py` to lock the new summary payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes a governance coverage summary without regressing the existing admin rollups.

影响范围

data-governance remediation from the V2.3 audit, limited to surfacing a baseline data dictionary and classification view for the core retained security, support, audit, and operational entities.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `data_governance_inventory` with entity classification, retention days, and purpose for task events, auth sessions, password recovery requests, support tickets, and retention audit records, and expanded `apps/api/tests/test_auth_wallet_tasks.py` to lock the new governance inventory payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now exposes a baseline data governance inventory without regressing the existing admin rollups.

影响范围

data-governance remediation from the V2.3 audit, limited to surfacing recent retention execution records directly inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `recent_retention_runs` with the latest retention execution IDs, request IDs, dry-run/confirm flags, deletion summaries, and timestamps, and expanded `apps/api/tests/test_auth_wallet_tasks.py` to seed retention executions and lock the new monitoring payload with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming retention execution history is now visible in the admin monitoring overview without regressing existing monitoring rollups.

影响范围

data-governance remediation from the V2.3 audit, limited to making the persisted retention execution records visible through an admin read path.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/routes/admin.py` so `/api/admin/retention/history` now returns recent persisted retention execution records with request IDs, dry-run/confirm flags, and deletion summaries, and expanded `apps/api/tests/test_retention_cleanup.py` to lock the new history query with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_cleanup.py -q` passed with 5 tests, confirming retention execution records can now be queried back in descending recency order without breaking the existing retention controls.

影响范围

data-governance remediation from the V2.3 audit, limited to storing each retention cleanup or dry-run execution as a durable audit record instead of relying only on logs.

这轮变更

extended `apps/api/app/models.py`, `apps/api/app/schemas/task.py`, `apps/api/app/services/tasks.py`, and `apps/api/app/routes/admin.py` so each `/api/admin/retention/run` call now writes a `retention_cleanup_executions` record with admin user, request ID, dry-run/confirm flags, and deletion counts, returns its `audit_record_id`, and expanded `apps/api/tests/test_retention_cleanup.py` to lock the persisted audit path with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_cleanup.py -q` passed with 4 tests, confirming retention runs now create durable execution records while preserving the existing delete, dry-run, confirmation, and audit-log behavior.

影响范围

data-governance remediation from the V2.3 audit, limited to preventing accidental destructive retention cleanup runs unless an admin explicitly confirms execution.

这轮变更

updated `apps/api/app/routes/admin.py` so `/api/admin/retention/run` now requires `confirm=true` for real deletes while still allowing `dry_run=true` previews, and expanded `apps/api/tests/test_retention_cleanup.py` to lock the destructive path, dry-run path, audit logging, and the new confirmation guard with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_cleanup.py -q` passed with 4 tests, confirming unconfirmed delete attempts are rejected and retention-eligible data stays untouched until `confirm=true` is supplied.

影响范围

data-governance remediation from the V2.3 audit, limited to recording structured audit logs whenever an admin runs retention cleanup or its dry-run preview.

这轮变更

updated `apps/api/app/routes/admin.py` so `/api/admin/retention/run` logs `admin_retention_cleanup_run` with the admin user, request ID, dry-run flag, and deleted-count summary, and expanded `apps/api/tests/test_retention_cleanup.py` to lock the new audit-log output with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_cleanup.py -q` passed with 3 tests, confirming retention cleanup still works and now emits the expected structured audit log.

影响范围

data-governance remediation from the V2.3 audit, limited to making admin retention cleanup previewable before it performs destructive deletes.

这轮变更

extended `apps/api/app/schemas/task.py`, `apps/api/app/services/tasks.py`, and `apps/api/app/routes/admin.py` so `/api/admin/retention/run?dry_run=true` returns the same retention policy and deletion counts without deleting records, and expanded `apps/api/tests/test_retention_cleanup.py` to lock both destructive and dry-run behavior with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_cleanup.py -q` passed with 2 tests, confirming dry-run returns cleanup counts while leaving the retention-eligible records untouched.

影响范围

data-governance remediation from the V2.3 audit, limited to turning the retention preview into an executable cleanup path for expired events, sessions, recovery requests, and closed support tickets.

这轮变更

extended `apps/api/app/schemas/task.py`, `apps/api/app/services/tasks.py`, and `apps/api/app/routes/admin.py` so `/api/admin/retention/run` now deletes records that exceed the configured retention windows and returns deleted counts, and added `apps/api/tests/test_retention_cleanup.py` to lock the cleanup path with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_cleanup.py -q` passed with 1 test, confirming the admin retention cleanup path deletes retention-eligible records and reports the deleted counts.

影响范围

data-governance remediation from the V2.3 audit, limited to surfacing retention policy windows and purge-eligible record counts inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/config.py`, `apps/api/app/schemas/task.py`, and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `retention_policy_days` and `retention_candidates`, and added `apps/api/tests/test_retention_preview.py` to lock task-event, auth-session, recovery-request, and support-ticket retention previews with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_retention_preview.py -q` passed with 1 test, confirming the monitoring overview now includes retention policy windows and purge-eligible record summaries.

影响范围

observability remediation from the V2.3 audit, limited to surfacing longer-window provider reliability, startup, and success-rate trend changes inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `provider_health_trend_deltas`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to lock the new 24h baseline vs 1h current-window provider health rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes provider health trend deltas without breaking the existing admin rollups.

影响范围

frontend smoke-gate remediation from the V2.3 audit, limited to bypassing Playwright's failing Chromium launcher on this Windows machine.

这轮变更

updated `apps/web/scripts/run-playwright-e2e.mjs` to detect a local Chrome/Edge binary, spawn it directly with a remote debugging port, and attempt `chromium.connectOverCDP()` instead of `browserType.launch()` for the smoke flow.

验证结果

`npm run test:e2e -- --project=chromium` now gets past the old `browserType.launch: spawn EPERM` failure and reaches a new boundary where the system browser never exposes its CDP endpoint in this environment, timing out on `http://127.0.0.1:9223/json/version`.

影响范围

observability remediation from the V2.3 audit, limited to surfacing short-window provider reliability, startup, and success-rate changes inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `provider_health_deltas`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed older and newer `degraded-cloud` snapshots and lock the new health-delta rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes provider health delta summaries without breaking the existing admin rollups.

影响范围

observability remediation from the V2.3 audit, limited to surfacing active retry backlog age by provider inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `retry_backlog_age_by_provider`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed an aged `runpod` retry backlog task and lock the new retry-aging rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes retry backlog aging summaries by provider.

影响范围

observability remediation from the V2.3 audit, limited to surfacing active cancellation and cleanup backlog age by provider inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `cleanup_backlog_age_by_provider`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed aged `runpod` and `vastai` backlog tasks and lock the new backlog-aging rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes cleanup backlog aging summaries by provider.

影响范围

observability remediation from the V2.3 audit, limited to surfacing recent cleanup failures split by provider and reason inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `cleanup_failures_24h_by_provider_reason`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed `runpod` and `vastai` cleanup failures with distinct reasons and lock the new cleanup provider+reason rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes provider-specific cleanup failure reason summaries.

影响范围

observability remediation from the V2.3 audit, limited to surfacing recent failed-task reasons split by provider inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `failed_tasks_24h_by_provider_reason`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed `runpod` and `vastai` failures with distinct reasons and lock the new provider+reason rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes provider-specific failed-task reason summaries.

影响范围

observability remediation from the V2.3 audit, limited to surfacing near-term failed-task reasons grouped by `Task.last_error` inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `failed_tasks_24h_by_reason`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed repeated failed tasks with the same error and lock the new failure-reason rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes repeated failed-task reason summaries.

影响范围

observability remediation from the V2.3 audit, limited to surfacing stuck active runs grouped by `TaskRun.status` inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `stale_active_runs_by_status`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed stale `executing` and `cleaning_up` runs and lock the new status rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes stale active-run summaries grouped by execution status.

影响范围

observability remediation from the V2.3 audit, limited to surfacing 24-hour actual-versus-quoted provider cost drift inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `provider_cost_drift_24h_by_provider`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed a finalized task cost drift and lock the new rollup with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes provider cost drift summaries.

影响范围

observability remediation from the V2.3 audit, limited to surfacing stale active runs and the oldest active run age inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `oldest_active_run_age_seconds` and `stale_active_runs_by_provider`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed a long-running active run and lock the new aging summaries with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes active run aging summaries.

影响范围

observability remediation from the V2.3 audit, limited to aggregating degraded provider snapshots and open cancellation backlog counts inside the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `degraded_providers` and `cancellation_backlog_by_provider`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed a degraded provider snapshot plus an open cancelling task and lock both summaries with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes degraded-provider and cancellation-backlog summaries.

影响范围

observability remediation from the V2.3 audit, limited to aggregating recent cleanup-related warning/error events and retried-task counts by provider in the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `cleanup_failures_24h_by_provider` and `retried_tasks_24h_by_provider`, and updated `apps/api/tests/test_auth_wallet_tasks.py` to seed retry and cleanup events and lock the new provider rollups with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes cleanup-failure and retried-task provider summaries.

影响范围

observability remediation from the V2.3 audit, limited to aggregating recent warning/error stages and failed-task provider counts in the existing admin monitoring overview.

这轮变更

extended `apps/api/app/schemas/task.py` and `apps/api/app/services/tasks.py` so `/api/admin/monitoring/overview` now returns `warning_events_24h_by_stage` and `failed_tasks_24h_by_provider`, and added regression coverage in `apps/api/tests/test_auth_wallet_tasks.py` to lock the new summaries against seeded warning/error events and a recent failed task.

验证结果

`python -m pytest apps/api/tests/test_auth_wallet_tasks.py -q` passed with 7 tests, confirming the monitoring overview now includes 24-hour warning-stage and failed-provider trend summaries.

影响范围

observability remediation from the V2.3 audit, limited to structured tracing inside provider marketplace service calls and worker cleanup/failure branches.

这轮变更

instrumented `apps/api/app/services/provider_marketplace.py` so offer fetch, submit, status poll, cancel, cleanup, and result collection all emit structured logs with the active `request_id`, and updated `apps/worker/worker/scheduler.py` to log cleanup start, cleanup finish, cleanup failure, and attempt failure with the same request context; added `apps/worker/tests/test_worker_cleanup_observability.py` for cleanup-path regression coverage.

验证结果

`python -m pytest apps/worker/tests/test_worker_cleanup_observability.py apps/api/tests/test_worker_request_trace.py apps/api/tests/test_business_observability.py apps/api/tests/test_observability.py -q` passed with 6 tests, confirming adapter and cleanup-path logs now keep the same request ID across cancellation and provider cleanup.

影响范围

observability remediation from the V2.3 audit, limited to persisting the task creation request ID and reusing it in worker execution and provider-dispatch logs.

这轮变更

added `request_id` storage on `Task` in `apps/api/app/models.py`, updated `apps/api/app/services/tasks.py` plus `apps/api/app/routes/tasks.py` so API task creation persists the inbound request ID, and instrumented `apps/worker/worker/scheduler.py` to bind that request ID back into the log context and emit structured worker lifecycle logs for task start, provider dispatch acceptance, result collection, and completion; added `apps/api/tests/test_worker_request_trace.py` for end-to-end regression coverage.

验证结果

`python -m pytest apps/api/tests/test_worker_request_trace.py apps/api/tests/test_business_observability.py apps/api/tests/test_observability.py -q` passed with 5 tests, confirming worker logs now carry the same request ID that entered at task creation.

影响范围

observability remediation from the V2.3 audit, limited to carrying `request_id` into core provider and task business logs without changing worker execution or external monitoring.

这轮变更

added `apps/api/app/services/observability.py` for request-id context and structured API event logging, updated `apps/api/app/main.py` to bind request IDs into request context, and instrumented `apps/api/app/routes/providers.py` plus `apps/api/app/routes/tasks.py` so quote creation, provider offer listing, and task creation emit structured logs with the active `request_id`; added `apps/api/tests/test_business_observability.py` for regression coverage.

验证结果

`python -m pytest apps/api/tests/test_business_observability.py apps/api/tests/test_observability.py -q` passed with 4 tests, confirming provider and task logs now include the same request IDs seen at the API boundary.

影响范围

observability remediation from the V2.3 audit, limited to backend request correlation and unhandled-exception logging without adding external monitoring vendors.

这轮变更

updated `apps/api/app/main.py` to attach `X-Request-ID` to every response, preserve inbound request IDs when present, and return structured 500 responses with matching request IDs while logging JSON error records for unhandled exceptions; added `apps/api/tests/test_observability.py` to verify both request ID headers and structured error logs.

验证结果

`python -m pytest apps/api/tests/test_observability.py -q` passed with 2 tests, confirming request ID propagation and structured exception logging work.

影响范围

legal and data-compliance remediation from the V2.3 audit, limited to a minimal site-wide cookie consent prompt with persistent user choice.

这轮变更

added `apps/web/src/components/cookie-consent-banner.tsx` to show accept-all and essential-only choices with links to `/privacy` and `/terms`, and mounted the banner globally from `apps/web/src/app/layout.tsx`.

验证结果

`npm run build` passed after the banner was added, confirming the site-wide consent UI compiles cleanly.

影响范围

account security remediation from the V2.3 audit, limited to backend TOTP enrollment, enable/disable, and login verification without taking on frontend UI in the same round.

这轮变更

added TOTP config in `apps/api/app/config.py`, added user-level TOTP fields in `apps/api/app/models.py`, added two-factor setup/status/enable/disable schemas in `apps/api/app/schemas/auth.py`, implemented TOTP secret generation and verification in `apps/api/app/services/security.py`, updated `apps/api/app/routes/auth.py` so login enforces `totp_code` when enabled and added `/api/auth/2fa`, `/api/auth/2fa/setup`, `/api/auth/2fa/enable`, and `/api/auth/2fa/disable`, and covered the flow in `apps/api/tests/test_two_factor_auth.py`.

验证结果

`python -m pytest apps/api/tests/test_two_factor_auth.py apps/api/tests/test_refresh_token.py apps/api/tests/test_session_management.py -q` passed with 7 tests, confirming TOTP setup, enable, login enforcement, disable, and existing refresh/session behavior all work together.

影响范围

account security remediation from the V2.3 audit, limited to backend device-session management for refresh-token sessions without taking on 2FA in the same round.

这轮变更

added `AuthSession` in `apps/api/app/models.py`, added session read/revoke schemas in `apps/api/app/schemas/auth.py`, updated `apps/api/app/routes/auth.py` so register/login/refresh persist and rotate refresh sessions, added `/api/auth/sessions` and `/api/auth/sessions/{session_id}`, and covered listing, rotation, and revocation in `apps/api/tests/test_session_management.py`.

验证结果

`python -m pytest apps/api/tests/test_session_management.py apps/api/tests/test_refresh_token.py -q` passed with 5 tests, confirming sessions are listed, refresh rotation revokes the old session, and a revoked session can no longer refresh tokens.

影响范围

user data-rights remediation from the V2.3 audit, limited to a password-confirmed DSAR delete flow for the signed-in user's repo-owned account data.

这轮变更

added `DeleteAccountRequest` and `AccountDeletionAccepted` in `apps/api/app/schemas/auth.py`, implemented `/api/auth/delete-account` plus owned-data cleanup for wallet, ledger, tasks, events, support tickets, projects, and password-recovery rows in `apps/api/app/routes/auth.py`, and added `apps/api/tests/test_account_delete.py` to cover successful deletion, token invalidation, cleanup, and wrong-password rejection.

验证结果

`python -m pytest apps/api/tests/test_account_delete.py -q` passed with 2 tests, confirming the account is deleted, owned rows are removed, the old token stops working, and the delete endpoint rejects an invalid current password.

影响范围

account security remediation from the V2.3 audit, limited to adding refresh-token issuance and token rotation in the auth API without changing the frontend flow yet.

这轮变更

added refresh-token expiry config in `apps/api/app/config.py`, added signed refresh-token helpers in `apps/api/app/services/security.py`, extended `AuthResponse` plus a new `RefreshTokenRequest` in `apps/api/app/schemas/auth.py`, updated register/login in `apps/api/app/routes/auth.py` to return both access and refresh tokens, added `/api/auth/refresh`, and covered the flow in `apps/api/tests/test_refresh_token.py`.

验证结果

`python -m pytest apps/api/tests/test_refresh_token.py -q` passed with 3 tests, confirming register now returns a refresh token, refresh rotation issues a new auth pair, and access tokens are rejected at the refresh endpoint.

影响范围

frontend smoke gate, limited to replacing the remaining `@playwright/test` worker-host execution path with a single-process Playwright browser script.

这轮变更

rewrote `apps/web/scripts/run-playwright-e2e.mjs` to execute the existing smoke flow directly with the `playwright` API after the repo-local API and web services are healthy, including support for `--project` and `--list`, instead of delegating to `@playwright/test/cli.js`.

验证结果

`npm run test:e2e -- --project=chromium` now gets through API startup, web build/start, and the full smoke runner bootstrap before failing later at `browserType.launch` with `spawn EPERM`, which narrows the remaining Windows blocker to Chromium process launch rather than Playwright test-worker orchestration.

影响范围

frontend smoke gate, limited to bypassing Playwright's `webServer` service-launch path by starting the API and web app from a repo-managed Node script before the smoke test begins.

这轮变更

added `apps/web/scripts/run-playwright-e2e.mjs` to start the existing API and web wrappers, wait for `/api/health` and `/login`, then invoke `@playwright/test/cli.js` with `PLAYWRIGHT_REUSE_EXISTING_SERVER=1`, and updated `apps/web/package.json` so `test:e2e` uses that runner.

验证结果

`npm run test:e2e -- --project=chromium` now gets through API startup, web build/start, and test discovery before failing later at `playwright/lib/runner/processHost.js` with `spawn EPERM`, which narrows the remaining Windows blocker to Playwright's own worker-process fork.

影响范围

frontend smoke gate, limited to the web-side Playwright `webServer` command that still used `npm run build && npm run start` through a shell on Windows.

这轮变更

added `apps/web/scripts/start-playwright-web.mjs` to run `next build` and `next start` sequentially via `process.execPath`, and updated `apps/web/playwright.config.ts` to use that wrapper instead of the shell-based `npm run build && npm run start` command.

验证结果

`npm run test:e2e -- --project=chromium` still fails immediately with `spawn EPERM`, which confirms the remaining Windows blocker is broader than the removed shell command chain.

影响范围

frontend release gate, limited to the Next.js TypeScript validation worker that was still failing with Windows `spawn EPERM` during production builds.

这轮变更

added `experimental.workerThreads = true` in `apps/web/next.config.ts` so Next.js runs its TypeScript validation worker through `worker_threads` instead of a spawned child process while keeping the single-worker build setting from the previous round.

验证结果

`npm run build` now passes end-to-end in `apps/web`, including `Running TypeScript`, `Collecting page data using 1 worker`, and static page generation.

影响范围

frontend release gate, limited to the repo-configurable Next.js worker count that was spawning 23 page-data workers on Windows.

这轮变更

added `experimental.cpus = 1` in `apps/web/next.config.ts` so Next.js uses a single worker for page-data collection instead of the previous multi-worker default.

验证结果

`npm run build` still fails at the existing TypeScript `spawn EPERM` stage, but `npx next build --experimental-build-mode compile` now reports `Collecting page data using 1 worker ...` before hitting the remaining Windows `spawn EPERM` child-process failure.

影响范围

frontend unit-test gate, limited to the remaining fixture typing mismatches inside `apps/web/src/app/tasks/[taskId]/page.test.tsx`.

这轮变更

added explicit fixture types for task, event, artifact, and detail objects in `apps/web/src/app/tasks/[taskId]/page.test.tsx`, then widened the shared test fixtures so nullable fields and array payloads stop inferring as `null`-only or `never[]`.

验证结果

`npx tsc --noEmit` now passes with no TypeScript errors in `apps/web`.

影响范围

frontend unit-test gate, limited to the TypeScript wiring for `vitest` globals and `@testing-library/jest-dom` matchers.

这轮变更

added explicit `node`, `vitest/globals`, and `@testing-library/jest-dom` types in `apps/web/tsconfig.json` and created `apps/web/vitest.d.ts` to load the Vitest jest-dom matcher declarations during TypeScript checking.

验证结果

`npx tsc --noEmit` now clears the large batch of matcher-type failures and stops only on 3 remaining real fixture typing errors in `src/app/tasks/[taskId]/page.test.tsx`.

影响范围

user data-rights remediation from the V2.3 audit, limited to giving authenticated users a machine-readable export of their own account, wallet, ticket, and recovery-request data.

这轮变更

implemented `/api/auth/export` in `apps/api/app/routes/auth.py` to return the signed-in user's account snapshot, wallet balances, task list, support tickets, and password recovery records, and added `apps/api/tests/test_account_export.py` to verify both authenticated export access and anonymous rejection.

验证结果

`python -m pytest apps/api/tests/test_account_export.py -q` passed with 2 tests, confirming authenticated users can export their account snapshot and unauthenticated requests are denied.

影响范围

account security hardening from the V2.3 audit, limited to letting authenticated users replace their own password safely.

这轮变更

added `ChangePasswordRequest` and `PasswordChangeAccepted` in `apps/api/app/schemas/auth.py`, implemented `/api/auth/change-password` in `apps/api/app/routes/auth.py` with current-password, minimum-length, and non-reuse checks, and added `apps/api/tests/test_change_password.py` to verify both successful rotation and rejection of a wrong current password.

验证结果

`python -m pytest apps/api/tests/test_change_password.py -q` passed with 2 tests, confirming the old password stops working and the new password works immediately after the change.

影响范围

API security hardening from the V2.3 audit, limited to abuse protection on register, login, and password-recovery requests.

这轮变更

added an in-memory rate-limit service in `apps/api/app/services/rate_limit.py`, enforced per-client per-email request limits in `apps/api/app/routes/auth.py`, reset limiter state in `apps/api/tests/conftest.py`, and added `apps/api/tests/test_rate_limit.py` to lock the 429 and `Retry-After` behavior.

验证结果

`python -m pytest apps/api/tests/test_rate_limit.py -q` passed with 2 tests, confirming auth abuse throttling now returns HTTP 429 with a retry window.

影响范围

frontend authentication storage hardening from the V2.3 audit, limited to the token and cached user session persistence layer.

这轮变更

updated `apps/web/src/lib/api.ts` so auth tokens and cached user data are written to `sessionStorage`, legacy `localStorage` values are migrated once into the session store, and logout now clears both stores.

验证结果

`npm run build` still reaches the existing late Next.js TypeScript `spawn EPERM` blocker after compile, and this auth-storage refactor introduced no earlier build-stage errors.

影响范围

authentication security baseline from the V2.3 audit, limited to the backend password hashing scheme and legacy-hash verification.

这轮变更

updated `apps/api/app/services/security.py` so new passwords are hashed with `bcrypt` while existing `pbkdf2_sha256` hashes remain valid, and added `apps/api/tests/test_security.py` to lock both behaviors with regression coverage.

验证结果

`python -m pytest apps/api/tests/test_security.py -q` passed with 2 tests, confirming bcrypt is now the default hash format and legacy pbkdf2 hashes still verify.

影响范围

web build reliability for the V2.3 test gate, limited to `next/font/google` blocking local builds and downstream smoke setup.

这轮变更

removed `next/font/google` usage from `apps/web/src/app/layout.tsx` and defined repo-local body and display font stacks via `--font-body` and `--font-display` in `apps/web/src/app/globals.css`.

验证结果

`npm run build` now compiles successfully instead of failing on Google Fonts fetches; the remaining build blocker has moved forward to a later Next.js TypeScript step that still ends with `spawn EPERM`.

影响范围

Playwright smoke runner setup, limited to the webServer launch path that still fails with `spawn EPERM` on this Windows machine.

这轮变更

added `apps/web/scripts/start-playwright-api.mjs` so the API can be launched through a repo-local Node wrapper, updated `apps/web/playwright.config.ts` to use that wrapper for the API server command, and made both webServer entries honor `PLAYWRIGHT_REUSE_EXISTING_SERVER=1` so local pre-started services can be reused instead of going through Playwright's failing process launcher.

验证结果

`npm run test:e2e -- --project=chromium` still fails before smoke execution because Playwright's own webServer launcher hits `spawn EPERM` when it attempts to start services itself; the repo now has a reusable-server path for the next verification round.

影响范围

Playwright smoke test stability, limited to the ambiguous homepage locator called out in the V2.3 gap audit.

这轮变更

narrowed the homepage brand assertion in `apps/web/tests/e2e/smoke.spec.ts` from a page-wide `/taskgpu/i` link lookup to a header-scoped brand home link check and added a single-match assertion.

验证结果

`npm run test:e2e -- --project=chromium` no longer hits the old ambiguous selector path in code, but this environment still fails earlier with a Playwright-level `spawn EPERM` before the smoke flow starts.

影响范围

frontend unit-test gate, limited to the Windows Vitest startup and worker configuration path.

这轮变更

converted the preload module passed to `node --import` into a file URL in `apps/web/scripts/run-vitest.mjs`, switched the default Vitest config loader to `native`, pinned Vitest to a single-thread `threads` pool in `apps/web/vitest.config.mjs`, and synchronized the Windows `net use` compatibility patch back into Node's ESM `child_process` exports in `apps/web/scripts/vitest-windows-safe-realpath.mjs`.

验证结果

`npm test` now loads all 13 frontend test files instead of failing immediately at startup; the remaining blocker is `vite:esbuild` failing with `spawn EPERM` during test transforms in this Windows environment.

影响范围

frontend security baseline, starting with response headers in Next.js.

这轮变更

added `Content-Security-Policy`, `Referrer-Policy`, `X-Content-Type-Options`, `X-Frame-Options`, `Permissions-Policy`, `Cross-Origin-Opener-Policy`, and `Strict-Transport-Security` in `apps/web/next.config.ts`.

验证结果

`npm run build` passed after the change and `/progress` remains reachable.