cm-extract-learnings

What problem does it solve?

Consolidates ad-hoc session notes and recurring signals into durable project memories so future sessions and teammates benefit from discovered patterns, gotchas, and configuration lessons without repeating work.

Core Features & Use Cases

• Parallel discovery and audit: Runs a Signal Discoverer to find new learnings and a Memory Auditor to verify and deduplicate against existing memories.
• Layered memory placement: Maps findings to the correct memory layer (global preferences, repo-level architecture, concise project memory, or long-form references) and enforces placement rules.
• Structured consolidation workflow: Performs orientation (resolve paths, read MEMORY.md and CLAUDE.md, capture git log), parallel agent gathering, synthesis with ranked proposals, and gated execution with line-count and global-file warnings.
• Use Case: Use when you want to "remember this" or run a "dream"/consolidation to prune outdated memories, promote recurring patterns, and keep the project memory under a practical size.

Quick Start

Tell the assistant to extract learnings from this project and propose memory additions, edits, or removals for consolidation.