The project brain for your codebase.

AI agents and humans should share one persistent understanding of a software project. Ruminate is that layer — agents capture and query through MCP; you inspect, curate, and steer in the web app.

MCP for agents · Web app for humans

How it works

  1. 01

    Connect a repository

    Static analysis maps every file, import, component and route — facts, not guesses. That part is free and takes seconds.

  2. 02

    Agents work through MCP

    Claude Code, Cursor or Codex reads and writes the project brain on the subscription you already pay for. Discoveries, systems, and concepts accumulate in one place.

  3. 03

    You steer from the app

    Overview, Knowledge, Architecture, Decisions, Questions, Activity — the human control plane for what your team and agents currently understand about the software.

“Finally — one place to see what our agents actually think they know about this codebase.”

Start with your own repository