✨ New
- Built-in Managed Investigation Agent with Intel Feed exposure checking
- Planning mode selector in agent creation UI
💎 Improvements
- Filter built-in agents from agent builder UI
- User email now included in application logs
- Google Drive auth now displays authenticated user
- Delete user button added to accounts page
- Cron-triggered agent run titles are now static
🐞 Fixes
- Fixed SentinelOne and agent system prompt issues
- Streamlined Google Drive authentication flow
- Fixed type errors in agent loop
- Removed
anytype usage for better type safety - Agent loop no longer throws unhandled errors
- Fixed missing tools in agent execution loop
✨ New
- Confluence documentation integration
- Agent performance metrics in logs and lists
- Scanner environment job with dedicated UI
💎 Improvements
- Added support for observables in theHive
- API documentation improvements
🐞 Fixes
- Fixed 400 error responses
- Removed LRU cache from tool credentials
- Fixed insights page default graph and tab
- Fixed documentation generation
💎 Improvements
- Linear status detection for better model understanding
- More aggressive context window management
✨ New
- LLM judge scoring in agent details UI
- Context manager implementation
💎 Improvements
- User deletion capability for admins
🐞 Fixes
- Fixed visual bugs in agent evaluation UI
- Fixed chat interruption error display
- Fixed Splunk and Hive schemas for Gemini/OpenAI
- Fixed time conversion tool for OpenAI