- Downgrade Python version from 3.13 to 3.11.12 for compatibility
- Improve empty LLM response handling with better diagnostics and retry logic in AnalysisAgent
- Add detailed logging for empty response retries with token count and iteration tracking
- Implement fallback result generation instead of immediate failure on consecutive empty responses
- Enhance stream error handling with partial content recovery and error message propagation
- Add comprehensive exception handling in stream_llm_call to prevent error suppression
- Implement project scope filtering to ensure consistent filtered views across Orchestrator and sub-agents
- Track filtered files and directories separately when target_files are specified
- Add scope_limited flag and scope_message to project structure for transparency
- Remove manual progress_percentage setting and rely on computed property for COMPLETED status
- Improve code comments with diagnostic markers (🔥) for critical sections
- Add database migrations for agent checkpoints and tree node tracking
- Implement core agent execution framework with executor, state management, and message handling
- Create knowledge base system with framework-specific modules (Django, FastAPI, Flask, Express, React, Supabase)
- Add vulnerability knowledge modules covering authentication, cryptography, injection, XSS, XXE, SSRF, path traversal, deserialization, and race conditions
- Introduce new agent tools: thinking tool, reporting tool, and agent-specific utilities
- Implement LLM memory compression and prompt caching for improved performance
- Add agent registry and persistence layer for checkpoint management
- Refactor agent implementations (analysis, recon, verification, orchestrator) with enhanced capabilities
- Remove legacy agent implementations (analysis_v2, react_agent)
- Update API endpoints for agent task creation and project management
- Add frontend components for agent task creation and enhanced audit UI
- Consolidate agent service architecture with improved separation of concerns
- This refactoring provides a scalable foundation for multi-agent collaboration with knowledge-driven decision making and state persistence
- Implement dual-mode streaming: prioritize in-memory EventManager for running tasks with thinking_token support
- Add fallback to database polling for completed tasks without thinking_token replay capability
- Introduce SSE event formatter utility for consistent event serialization across streaming modes
- Add 10ms micro-delay for thinking_token events to ensure proper TCP packet separation and frontend incremental rendering
- Refactor stream_agent_with_thinking endpoint to support both runtime and historical event streaming
- Update event filtering logic to handle both in-memory and database event sources
- Improve logging with debug markers for thinking_token tracking and stream mode selection
- Optimize polling intervals: 0.3s for running tasks, 2.0s for completed tasks
- Reduce idle timeout from 10 minutes to 1 minute for completed task streams
- Update frontend useAgentStream hook to handle unified event format from dual-mode streaming
- Enhance AgentAudit UI to properly display streamed events from both sources
- Introduce LLM-driven decision-making across various agents, allowing for dynamic adjustments based on real-time analysis.
- Implement new event types for LLM thinking, decisions, actions, and observations to enrich the event streaming experience.
- Update agent task responses to include additional metrics for better tracking of task progress and outcomes.
- Refactor UI components to highlight LLM-related events and improve user interaction during audits.
- Enhance API endpoints to support new event structures and improve overall error handling.
- Introduce new Agent audit functionality for autonomous code security analysis and vulnerability verification.
- Add API endpoints for managing Agent tasks and configurations.
- Implement UI components for Agent mode selection and embedding model configuration.
- Enhance the overall architecture with a focus on RAG (Retrieval-Augmented Generation) for improved code semantic search.
- Create a sandbox environment for secure execution of vulnerability tests.
- Update documentation to include details on the new Agent audit features and usage instructions.
google-labs-jules[bot]
lintsinghua