reflect

Overview: The Reflect skill is a self-improvement tool that helps agents learn from conversations and improve their performance. It analyzes corrections, success patterns, and user feedback to propose updates to agent files and create new skills. Key Features: * Analyzes conversations for learnings and proposes updates to agent files * Creates new skills based on reusable patterns and insights * Integrates with state manager and signal detector for efficient learning * Supports multiple learning categories and target files How It Works: The Reflect skill follows a five-step workflow: 1. Initialize state files using the state manager 2. Scan conversation for signals using the signal detector 3. Classify and match signals to target files 4. Check for skill-worthy signals and propose new skills 5. Generate proposals for agent updates and new skills Use Cases: * Improve agent performance by learning from corrections and success patterns * Create new skills based on reusable patterns and insights * Enhance state management and signal detection for efficient learning