Manages and evolves project-specific knowledge in CLAUDE.md to improve AI performance over time.
Compound Learning is a specialized skill for Claude Code designed to build a persistent project memory by documenting technical patterns, successes, and failures in a standardized CLAUDE.md file. By capturing dated, actionable, and specific learnings after each development cycle, it ensures that Claude becomes increasingly context-aware and effective within a specific codebase. This system helps prevent recurring bugs, preserves architectural decisions, and maintains a high-signal documentation layer that scales as your project grows.
Key Features
01Automated project memory updates via the /ct:compound command
02Scalable architecture for splitting large documentation into topic-specific files
03Standardized entry formatting for dated, specific, and actionable insights
04Structured organization for effective patterns, failed approaches, and recurring bugs
050 GitHub stars
06Intelligent maintenance including deduplication and contradiction resolution
Use Cases
01Recording complex library configurations to prevent future integration regressions
02Tracking failed technical experiments to avoid repeating unproductive development paths
03Documenting architectural patterns and coding standards for team-wide consistency
