| name | verify-backend |
| description | Run and fix backend pytest tests after API, scanner, model, or router changes. Use when modifying backend/app/, adding endpoints, or when API tests fail. |
Verify backend changes
When to use
After any change under backend/app/ — routers, models, scanner, config, VLC/thumbnails.
Steps
- Run API tests:
.\scripts\test.ps1 -Layer api
- If failures occur, read the pytest output carefully
- Fix the smallest issue; re-run only the failing file:
cd backend
$env:APP_ENV="test"; $env:TEST_MODE="true"
.\.venv\Scripts\pytest tests/test_library_api.py -q
- Repeat until green
Test file map
| File | Covers |
|---|
test_health.py | GET /api/health, GET /api/thumbnails/status, POST /api/scan modes, GET /api/scan/status |
test_thumbnail_worker.py | thumbnail queue worker_status() |
test_folder_picker.py | Windows folder picker subprocess |
test_media_roots_api.py | media roots CRUD, pick-folder responses |
test_library_api.py | browse, movies, shows, show detail, search |
test_watch_api.py | watch-status, season watch-status, continue-watching |
test_play_api.py | play endpoint, TEST_MODE VLC skip |
test_settings_api.py | GET/PATCH settings, POST /api/settings/reset-data |
test_thumbnails.py | thumbnail cache, seek logic, cache version |
test_ffmpeg_detection.py | ffmpeg path detection, winget fallbacks |
test_scanner.py | scan_library against fixture media (see verify-scanner skill) |
Self-correction loop
run pytest → read traceback → fix code → re-run failing file → run full api layer
Notes
- Tests use isolated SQLite via
tmp_path; never touch backend/data/library.db
TEST_MODE=true skips VLC and thumbnail background jobs- Import
app.db as db and use db.engine after overrides (not from app.db import engine)
