- Add WebSocket proxy for /ws/task-progress in botui (was missing, blocking progress events)
- Fix task detail endpoint to use UUID binding for auto_tasks query
- Fix task list to query auto_tasks table instead of tasks table
- Add proper CSS for task cards with status-based colors
- Add task detail panel CSS for header, progress, sections
- Add count-all to stats HTML response
- Skip 0-byte files in drive monitor and document processor
- Add detailed logging for LLM calls in intent classifier and app generator
- Remove unused variables and fake demo activity simulation
- Change DRIVE_MONITOR checking logs from info to trace
- Fix hardcoded 'gpt-4' model in auto_task modules (intent_classifier, app_generator, designer_ai, intent_compiler) to use configured llm-model from bot config
- Add vector_db (Qdrant) to required bootstrap components for KB indexing
- Add Qdrant health check with clear error messages when unavailable
- Change verbose [START] debug messages from info to trace level
- Fix episodic memory role handling in Claude client (convert 'episodic' to system context)
- Disable auth for /api routes during development
- Add DynamicLLMProvider wrapper for runtime LLM provider updates
- Start DriveMonitor for default.gbai bucket on server startup
- Fix DriveMonitor to detect config.csv changes and update LLM provider
- Fix path matching to detect config.csv in root and .gbot folders
- Add /api/auth to anonymous paths to fix 401 on auth endpoint
- Fix foreign key references in 6.1.0 migration (users.id not users.user_id)
- Use correct bucket name 'default.gbai' for DriveMonitor
- Add json-enc extension to quick intent button in tasks.html
- Add json-enc extension to intent form in autotask.html
- Server expects JSON but HTMX was sending form-urlencoded data
- Fix pricing_model -> model field name in whatsapp mock
- Fix trivially_copy_pass_by_ref in BrowserType::browser_name
- Add methods to use Locator and Action enum fields (dead_code fix)
- Convert url_pattern methods to associated functions
- Replace panic! with bail! in e2e tests
- Collapse nested if statements, fix cmp_owned
- Remove redundant clones and needless borrows
- Add underscores to long numeric literals
- Replace assert!(true) with meaningful tests
- Remove unused imports, macros, and functions
- Fix async function without await
- Rename struct fields to avoid common suffix warnings
All clippy warnings and errors resolved: 0 warnings, 0 errors
