Learning Agent Pro

Learning Agent Pro is a comprehensive system that captures learnings, errors, and corrections to enable continuous improvement for your AI agents. Never lose a lesson twice.

🎯 WHAT YOU GET: • Automated error logging with full context • Correction tracking (user corrections become learnings) • Feature request capture • Knowledge gap identification • Pattern detection for recurring issues • Automatic promotion to workspace files (SOUL.md, AGENTS.md, TOOLS.md) • Optional hook for automatic session reminders • Inter-session communication for sharing learnings

🚀 HOW TO USE:

1. Download from ClawMart and extract to your OpenClaw skills folder
2. Run: mkdir -p ~/.openclaw/workspace/.learnings
3. The skill automatically logs errors, corrections, and feature requests
4. Check .learnings/LEARNINGS.md and .learnings/ERRORS.md for captured data
5. Promoted learnings appear in SOUL.md, AGENTS.md, and TOOLS.md

📊 LOGGING INCLUDES: • Error entries: command, output, context, suggested fix • Learning entries: priority, category, suggested action, source • Feature requests: user requests, capabilities needed • Timestamps, tags, related files, pattern keys

💡 USE CASES: • Coding agents that learn from their mistakes • Preventing repeat errors in long-running sessions • Building institutional knowledge across sessions • Tracking user corrections for improvement • Capturing feature requests for roadmap

📦 TECHNICAL DETAILS: • Works with OpenClaw, Claude Code, Codex, Copilot • Creates .learnings/ directory structure • Includes templates for errors, learnings, feature requests • Hook system for automatic reminders • Cross-session learning sharing

Perfect for: Anyone running AI agents who wants them to improve over time instead of repeating the same mistakes.