self-improving-agent

To promote Claude Code auto-memory, analyze MEMORY.md and topic files to identify actionable patterns, then consolidate them into CLAUDE.md or .claude/rules/ for durable project guidance. This extracts high-value learnings into permanent rules.

Auto-memory analysis reviews MEMORY.md and topic files to detect promotion candidates, stale entries, and consolidation opportunities. It cross-references CLAUDE.md and existing rules to eliminate duplicates, resolve conflicts, and extract reusable skills.

Clean up stale entries in MEMORY.md by running an auto-memory review that identifies outdated information and consolidation opportunities. The analysis cross-references CLAUDE.md to detect conflicts and gaps, guiding you to remove or update obsolete content.

Yes, you can extract reusable skills from Claude Code memory files by analyzing auto-memory to identify promotion candidates. The process surfaces patterns from MEMORY.md that can be consolidated and promoted to enable durable guidance across multiple projects.

Find duplicate rules in CLAUDE.md and .claude/rules/ by cross-referencing existing rules against auto-memory entries. The analysis detects duplicates, conflicts, and gaps, providing guidance for consolidating similar entries into a single source of truth.

Consolidate auto-memory topic files when analysis detects stale entries, duplicates, or overlapping patterns across MEMORY.md. The review process identifies consolidation opportunities and promotes durable learnings to CLAUDE.md, reducing fragmentation in your project rules.