Karpathy posted his raw/ folder workflow for LLM context — a simple… — Data1984

Data1984@data1984 · 775 subscribers Open in TelegramOpen

Karpathy posted his raw/ folder workflow for LLM context — a simple idea: dump your notes, PDFs, papers into one folder and feed it to the LLM.
Someone built a full system on top of it for Claude Code.

claude-memory-compiler hooks into your Claude Code sessions and automatically compiles them into a structured knowledge base. Every decision, pattern, lesson — extracted and cross-referenced into markdown articles.

The architecture is interesting:
• No RAG, no vector DB — just a smart index.md the LLM navigates
• Works at personal scale (50–500 articles) better than cosine similarity
• Runs on your existing Claude subscription, no extra API costs
https://github.com/coleam00/claude-memory-compiler

GitHub

GitHub - coleam00/claude-memory-compiler: Give Claude Code a memory that evolves with your codebase. Hooks automatically capture…

Give Claude Code a memory that evolves with your codebase. Hooks automatically capture sessions, the Claude Agent SDK extracts key decisions and lessons, and an LLM compiler organizes everything in...

t.me/data1984/960