Claude Memory Compiler — Evolving Knowledge Base

Introduction

Claude Memory Compiler is a personal knowledge base tool for Claude Code, with 365+ GitHub stars. It automatically captures Claude Code sessions via hooks, uses the Claude Agent SDK to extract key decisions and lessons learned, and compiles them into structured, cross-referenced knowledge articles. No vector database — it uses a structured index.md approach. Inspired by Karpathy's LLM knowledge base architecture. Ideal for heavy Claude Code users.

Claude Memory Compiler — Build Memory from AI Sessions

Workflow

1. Session ends → hooks trigger flush.py
2. Knowledge extraction → Claude Agent SDK identifies decisions, patterns, lessons
3. Daily log → saved to timestamped files
4. Compilation → compile.py organizes into cross-referenced articles
5. Injection → SessionStart hook loads index.md into the next session

Core Features

• Auto-capture — hooks trigger at session end
• No vector DB required — structured index.md works better at personal scale
• Health checks — lint.py runs 7 checks
• Query interface — command-line search of the knowledge base

FAQ

Q: What is Claude Memory Compiler? A: An open-source tool that auto-captures Claude Code sessions and compiles them into a structured, searchable knowledge base.

Q: Do I need a vector database? A: No — the structured index.md approach outperforms RAG at personal scale.