Problem 1: Remove legacy role bypass
- Remove role="admin" string bypass in permissions.py resolve_permissions()
- Remove role="admin"/"editor" bypass in auth.py check_permission()
- Remove legacy role string fallback in deps.py require_admin/require_write
- Add migration 0112: Create Role records for built-in roles and link role_id
- KI-Kommentar: Legacy Role Bypass entfernt — alle Admins müssen echte role_id haben
Problem 2: Enforce API token scopes
- Add _token_scopes check in require_permission() in deps.py
- When _token_scopes is set (API token auth), required permission must be in scopes
- When _token_scopes not set (session auth), normal permission check applies
Problem 3: Migration chain verification
- Chain is already linear: 0027→0028_rls_force→0028_user_preferences→0029
- user_preferences table confirmed exists in DB
- No duplicate revision IDs found
Problem 4: RLS for remaining tenant tables
- Add migration 0111: Dynamic RLS activation for any remaining tables with tenant_id
- Login tables and global tables explicitly excluded
- DB check shows 0 tables currently missing RLS (safety net migration)
Problem 5: Permission cache invalidation on tenant switch
- Add invalidate_permission_cache() call in switch_tenant() for old tenant
- Stale cached permissions from old tenant no longer leak
Problem 6+7: Guest system removal
- Remove get_current_guest() from deps.py
- Remove guest_auth.py router from main.py and routes/__init__.py
- Rewrite guests.py to use regular User/UserTenant with role=guest
- Remove GuestUser/GuestInvitation from models/__init__.py
- Add migration 0113: Migrate guest_users to regular users, drop guest tables
- Update frontend GuestLogin/GuestContacts to redirect to normal pages
- KI-Kommentar: Guest-System umgebaut — Guests sind jetzt reguläre User mit role=guest
Plugin route registration in main.py was mutating module-level router
singletons by appending require_active_plugin dependencies directly to
router.routes. This persisted across app instances, causing test routes
to inherit require_active_plugin checks and return 403 "plugin inactive"
when tests created their own FastAPI apps with those routers.
Fix: use app.include_router(router, dependencies=[plugin_dep]) which
adds dependencies at the app level without modifying the shared router.
Fixes 35 test failures across 4 test files:
- test_agent_memory.py (6 failures)
- test_external_agent_api.py (15 failures)
- test_graph_rag.py (7 failures)
- test_marketplace.py (7 failures)
- Use migration engine (crm_migration, BYPASSRLS) for default data seeding
in app/main.py instead of crm_api role which is RLS-enforced
- Skip domains PATCH for dockercompose apps in deploy_api() to avoid 422
- Regenerate migration_hashes.txt for 0085_restore_tenant_rls.py
- Add migration 0110: password_salt column to mail_accounts
- app/core/resilience.py: CircuitBreaker (CLOSED/OPEN/HALF_OPEN), retry_db,
redis_call_with_fallback, InMemoryRateLimiter, CircuitBreakerMiddleware
- app/core/auth.py: get_session_data now falls back to PostgreSQL sessions
table when Redis is unavailable
- app/core/permissions.py: get_cached_permissions falls back to direct DB
resolution when Redis circuit is open
- app/core/rate_limit.py: check_rate_limit falls back to in-memory limiter
when Redis is down; reset_rate_limit clears both Redis and in-memory
- app/core/middleware.py: CSRF validation uses get_session_data (Redis+DB
fallback); sliding session TTL is best-effort during outage
- app/core/db/__init__.py: get_db() wraps session creation with retry_db
for transient connection errors; records circuit breaker success/failure
- app/deps.py: refresh_session_ttl wrapped in try/except for Redis outage
- app/main.py: CircuitBreakerMiddleware registered (returns 503 when DB
circuit is OPEN, skips health/metrics endpoints)
- app/config.py: Added resilience settings (thresholds, cooldown, retries)
- tests/test_resilience.py: 30 tests covering all patterns
30/30 resilience tests pass. No regressions in plugin lifecycle tests.
- main.py: registry.initialize(get_migration_engine()) instead of get_engine()
- main.py: plugin migrations run via get_migration_session_factory() not async_session()
- registry.py: upgrade_plugin, install_plugin, uninstall_plugin all use migration session for DDL
- db/__init__.py: get_migration_engine() raises RuntimeError if MIGRATION_DATABASE_URL missing (no fallback)
- Fixes fresh-install failure: crm_api has no DDL rights, plugin migrations need crm_migration
wrap_plugin_route copied __signature__ from the original handler but
the wrapper's __globals__ namespace (plugin_error_handler.py) did not
contain the Pydantic models (ConversationCreate, MessageCreate, etc.).
FastAPI could not resolve ForwardRef('ConversationCreate') → 422 on
all POST routes with body parameters.
Removing the wrapper entirely fixes this. Plugin error isolation can
be re-added later using a different approach (middleware or exception handler).
Router-level dependencies=[Depends(require_active_plugin)] was applied
to ALL routes including WebSocket. Now adding the dependency per-HTTP-route
only, WebSocket routes are skipped entirely.
Simpler approach: require_active_plugin._check() now accepts Request
parameter and returns early for WebSocket upgrade requests.
No route splitting needed — all routes stay in their original router.
WebSocket routes were getting require_active_plugin dependency applied
via include_router(dependencies=[...]) which caused 403 Forbidden
before the WebSocket upgrade could happen.
Fix: Split router into HTTP routes (with dependency) and WebSocket routes
(registered separately without the active-plugin check). WebSocket auth
is handled inside the endpoint itself via session cookie verification.
1. plugin_error_handler.py: Remove _UploadFile alias, import UploadFile directly
so FastAPI can resolve ForwardRef('UploadFile') in the wrapper's namespace.
Also import WebSocket for ForwardRef resolution.
2. main.py: Skip WebSocket routes in wrap_plugin_route — WebSocket endpoints
must not be wrapped (different protocol, no JSONResponse on error)
- main.py: RequestLoggingMiddleware reports 4xx/5xx responses and unhandled exceptions to Forgejo
- worker.py: Plugin activation failures and outbox job failures reported to Forgejo
- 401/403 are NOT reported (expected auth/permission behavior)
- All other errors (422, 404, 500, network, worker) ARE reported
Agent Zero