The unique index on alert_rules.name was being created before duplicates
were cleaned up, causing DuplicateKeyError on startup when existing
duplicates were present. Move deduplication into setup_indexes() so it
runs before the unique index is created.
v1.7.6
- Add unique index on alert_rules.name in setup_indexes()
- seed_default_rules() now removes duplicates by name before upserting
- Keeps the oldest document (_id ascending) when deduping
- Add saved_searches_collection to database.py with index on created_by+created_at
- New routes/saved_searches.py: GET /api/saved-searches, POST, DELETE
- Saved searches are scoped per user (created_by = token sub)
- Mount router in main.py
- Frontend: Save filters button, saved search pills with load/delete
- loadSavedSearches called on initApp
- applySavedSearch restores filters and validates services against current options
- Add CSS for saved-searches row
- Add tests for CRUD, delete 404, and name validation
- Cache Graph API tokens with expiry-aware reuse in graph/auth.py
- Add tenacity-based retry/backoff wrapper (utils/http.py) and apply to all Graph/source API calls
- Add Pydantic request/response models (models/api.py) and FastAPI query constraints
- Add unit tests for event_model, auth and integration tests for API endpoints
- Configure ruff linter/formatter in pyproject.toml
- Add GitHub Actions CI pipeline (.github/workflows/ci.yml)
- Add requirements-dev.txt with pytest, mongomock, httpx, ruff
- Clean up typing imports and fix ruff linting across codebase
