# memtomem

A markdown-first, long-term memory tool for AI coding agents that indexes your notes, documents, and code with hybrid keyword and semantic search, exposed to any agent through MCP.

memtomem turns markdown notes, documents, and code into a searchable knowledge base that any AI coding agent can query. Notes are written as plain `.md` files that stay the source of truth; memtomem indexes them into a rebuildable database and makes them searchable by both exact keywords (BM25) and meaning (semantic embedding), covering markdown, JSON, YAML, Python, and JavaScript/TypeScript files.

Core memory operations run only when explicitly called, and optional client hooks that automate lookups are explicit and removable rather than hidden automation. It is installed with `uv tool install` or `pipx`, with an `[all]` extra that bundles ONNX dense embeddings, a Korean tokenizer, Ollama and OpenAI provider support, a code chunker, and a web UI, on top of a smaller BM25-only install.

The project is alpha software, with APIs, defaults, and on-disk configuration still subject to change between 0.x releases.

Website: https://memtomem.com