fix: stabilize integration tests and improve agent drilldown UI compatibility

Refactor configuration flow (Day 0 vs Day 1 separation) and add startup scripts ...

- Stop AI provider settings from syncing back to config.yaml (stored in DB only now).
- Parameterize frontend port in docker-compose.
- Add start_server.sh and stop_server.sh scripts.
- Update .gitignore to exclude scratch files.

Security & Logic: Implemented ownership-based permission control for agents and added model_name override support for agent tasks

UI Fix (Hardened): Final scroll fix for ChatTracker using absolute positioning and triggering a full frontend rebuild to ensure effect

UI Fix: Enable scrolling in the ChatTracker by correctly constraining ChatWindow height

UI Polish: Remove redundant and non-functional Refresh Context button from Evaluation History

UI Fix: Prevent Evaluation sections from collapsing and add minimum heights to metrics cards for better layout stability

Fix: Implement cumulative metrics tracking in agent loop, fix metrics reset endpoint, and restore full metrics dashboard UI

Fix: Add missing handlers to useAgentDrillDown hook to resolve build errors

UI: Add ReactMarkdown to EvaluationPanel for beautiful feedback rendering

Refactor AgentDrillDown into modular components and fix evaluation feedback issues

fix(prod): restore stability and enhance agent observability ...

- fix: correct build context paths in docker-compose.test-nodes.yml for remote hosts
- fix: provide default 'skill_type' in system skill bootstrap to avoid IntegrityError
- fix: relax X-Proxy-Secret check for OIDC environments to allow UI steering
- feat: redesign agent status bar for high-density observability (Current vs Last Action)
- fix: correct .coworker.md path resolution in Evaluation Hub
- chore: remove obsolete 'Open Evaluation Hub' button

feat: stabilize browser streaming, fix quality audit UI readability, and harden integration test harness

Hardening Swarm Strategy: Unified .gitignore and Generic Manifesto

feat: comprehensive mobile responsiveness for agent dashboard and swarm control

feat: complete Co-Worker integration tests and finalize project roadmap

feat: implement Co-Worker DB models, API schemas, and UI Dashboard components

fix(metrics): fix agent tool metrics persistence and improve drilldown UI responsiveness

refactor: move SwarmControl File Explorer to a right sidebar on desktop

feat: Add one-click webhook copy to Dashboard cards

feat: Add Dual-Mode (Sync/Async) Webhook Support

docs: Add rich documentation tooltip for webhook URLs

feat: Enhance agent webhook with custom prompt support and UI fire button

Fix integration tests for agents, file sync, and audio. Restore 2026 model versions.

feat: add drilldown link icon indicator to agent dashboard cards

fix: resolve tool error counting logic and add advanced UI execution metrics

feat(agents): full implementation of harness engineering and file sync fixes ...

- Implemented Area 1: Backend scaffolding, CRUD, schemas, and session locking
- Implemented Area 2: Agent execution loop, RAG integration, and Circuit Breaker
- Updated UI: Configure CRON triggers, Target Mesh nodes, and display system prompt in Agent Drilldown
- Fixed LLM fallback resolving vertex_ai_beta
- Fixed File System Sync pseudo-nodes (hub, local) raising 'Offline'
- Fixed FileSystemNavigator initialPath bug causing synced files not to render in UI Mesh Workspace

fix(tool_service): fix LazyFileContent property indentation bug causing NoneType error

fix(frontend): corrected mesh node field mapping in AgentDrillDown ...

The Target Mesh Node dropdown was incorrectly using 'id', 'name', and 'status' instead of the backend's 'node_id', 'display_name', and 'last_status', causing nodes to appear as '(unknown)'.

fix(llm): eradicate gemini-1.5-flash 404 from production ...

- Implemented substring normalization for Gemini models in config.py
- Forced redirection of 1.5-flash to 2.5-flash in RAG service
- Fixed model name resolution in provider factory to handle suffixed names
- Normalized health checks to use 2.5-flash baseline
- Removed hardcoded 1.5-flash defaults from STT/TTS providers and agent loop

Fix frontend build crash by updating getNodes to getUserAccessibleNodes and restore production service sync