Give your agent a memory it can't lose

Coding OS is an open-source cognitive operating system for AI coding agents. It installs on top of the agent you already use — Claude Code or Codex — and adds memory, a knowledge graph, and an enforced engineering loop.

What you get

graph_os

A precomputed knowledge graph of your code, docs, and configs. The agent asks “who calls this / what breaks” in one query instead of grepping the repo.

thinking_os

Cross-session memory, learning, and metrics. Every session records what it learned; the next one starts with a digest instead of amnesia.

board_os

A Scrumban board with WIP limits, a definition of ready, and lifecycle hooks. Tasks are contracts the agent has to satisfy — not vibes.

Start here

Installation

Requirements, install the cos CLI, and initialize your first project.

Quickstart

Five minutes from install to an agent with memory and a task board.

How it works

The three brains, the complexity gate, and the loop hooks enforce.

Adapters & support

Exactly what works on Claude Code, Codex CLI, Codex app, and for humans.

The Reference section is synced from the tool repository on every deploy — it is the contributor-facing engineering, architecture, and playbook documentation. Start with the guides above; reach for reference when you are extending Coding OS itself.