v2.119.0
7 days ago by Ethan Lin
New Features
- Sample Query Generation: Added comprehensive sample query generation feature with configurable maxTokens and maxTables options for database documentation
- Testing UI Package: Implemented new dialog-based test execution system with context-aware preselected modes and comprehensive testing infrastructure
- Hierarchical Agent Step Tracking: Added hierarchical step count system (e.g., 'Step 2.1.3') that shows full execution path for agents running in MJ Conversations
- Claude Code Commands: Added changeset and PR creation commands for streamlined development workflows
Improvements
- AI Prompt Engine: Enhanced effort level control with 1-100 integer scale and provider-specific mapping (OpenAI, Anthropic, Groq, Gemini)
- Agent Conversation UX: Improved conversation UI with consistent step counts, agent-specific icons in active tasks, and better button positioning
- CodeGen Enhancements: Enhanced advanced generation with category preservation and entity importance analysis
- Configuration Validation: Fixed config validation errors for commands that don't need database connections
- Virtual Fields Support: Fixed missing virtual fields and improved @mention chips for agents with config preset names
Bug Fixes
- Angular Change Detection: Fixed ExpressionChangedAfterItHasBeenCheckedError in Angular components
- Step Count Display: Fixed subagent step count consistency - no more disappearing or confusing jumps
- Skip SDK Integration: Fixed Skip SDK to include entity mappings in query metadata
- Package Documentation: Updated DBAutoDoc documentation and added comprehensive new package setup guide