Coordigent docs

Coordigent makes the coding agents working in the same repo aware of each other, and shows their work to the people running them.

#What it does

When several agents work in one repo at once — Claude Code, Codex, Cursor, Kiro and others — each is blind to the rest. Two edit the same file, one changes what another depends on, and you find out at merge time. Coordigent records what every agent is doing and which files it is in. An agent about to edit a file another agent is working in is warned first. Agents can send each other messages, an agent that is stuck shows up under Needs you, and an agent that runs out of context leaves a handoff for the next one.

It works on one machine with no account. Logged in, your whole project sees the same picture: every teammate's agents appear live in the terminal, in VS Code and on the website, and conflicts between teammates' agents reach both agents. Coordigent never blocks an edit, never pushes, and never uploads a transcript.

#Set up

Two commands, once per machine. A third is optional.

After coord login, agents in any repo on this machine can coordinate. The first agent in a new repo raises one question, whether to connect it to your project, and your answer decides whether your teammates see that repo's agents. Restart your agent app and start working. Install walks through each step.

#Where to go next

• Get started: Install, Log in, Connect a repo, Invite teammates and What you'll see.
• Tools: the CLI and its live view, the VS Code extension, this machine (the daemon and the local API), and the website.
• Concepts: projects and repos, agents, files in use, conflicts, Needs you, messages, handoffs and runs.
• Integrations: Claude Code, Codex, Cursor, Kiro, Windsurf, VS Code and Copilot, other clients and Slack.
• Reference: every command, MCP tools, the local API and headless and CI.

How your data is handled is on the security and privacy pages.