#135 · Primary category: Knowledge Base & RAG

cavemem

ai caveman claude claude-code compress memory rag rag-chatbot

Frozen — cross-agent persistent memory for coding assistants. Still works; the compressed-memory core now ships inside JuliusBrussee/caveman.

Project last updated:08/14/26

GitHub Stars

674

Forks

62

Contributors

6

License

MIT

Why we included this project

Anyone who runs a terminal-based coding agent, say Claude Code or Codex, and keeps re-explaining the same project context every session will recognize the problem this solves. cavemem hooks into the assistant at session boundaries, compresses what happened into a deterministic shorthand that keeps code, paths, and identifiers byte-for-byte while cutting prose tokens by roughly three quarters, and stores the result in local SQLite. The agent can then query its own history through three MCP tools, with hybrid keyword and vector search behind them and a ranker you can tune. Everything stays on your machine unless you deliberately enable a remote embedding provider, and a read-only web viewer at localhost:37777 makes past sessions browsable in plain language. The project is frozen as of August 2026, though it still installs and works; the compressed-memory core now lives on in the actively developed caveman repository.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category