Automates the curation of Claude Code's auto-memory to transform temporary insights into durable project rules and reusable skills.
The Self-Improving Agent acts as an intelligence layer for Claude Code’s automatic memory system, preventing critical project insights from being lost or truncated. By analyzing the MEMORY.md file, it identifies recurring patterns and 'graduates' them into enforced project standards within CLAUDE.md or scoped rules in .claude/rules/. This skill ensures your AI collaborator doesn't just passively record information, but actively improves its own performance and adherence to project-specific requirements through intelligent curation and skill extraction.
主要功能
01Analyzes MEMORY.md to identify promotion-ready patterns and stale entries
02Graduates background insights into enforced project rules (CLAUDE.md)
03Captures and structures command-line errors to improve future debugging accuracy
04Extracts recurring solutions into standalone, reusable skills with full documentation
05Provides a memory health dashboard to monitor line counts and capacity
060 GitHub stars
使用场景
01Converting successful debugging patterns into permanent project standards
02Migrating project-specific learnings into shareable skills for broader use
03Reviewing and consolidating what Claude has learned about a complex codebase
