Claude Code + Cursor
Use Claude Code and Cursor on the same project
Run Claude Code in the terminal and Cursor in the editor on one repository without each tool undoing the other's work.
The problem
Cursor's agent and Claude Code each read the repo from scratch. Neither knows the other is mid-change, and the two follow different instruction files.
What people try
- Using one tool at a time
- Safe, but you lose the reason you have both.
- Copying the same rules into .cursor/rules and CLAUDE.md
- Keeps standing rules aligned, but not who is working on what right now.
With Concord
01
Set up once
concord setup registers the MCP server in .mcp.json and .cursor/mcp.json and writes Concord instructions into CLAUDE.md and .cursor/rules/.
02
Claim from either side
Whichever tool starts a task claims it, and the other sees the claim and any overlap before it edits.
03
Message across
Either agent can prompt the other. Cursor delivery depends on its background monitor, and concord adapters status shows whether it is reachable.
Result
The editor agent and the terminal agent share one picture of the work.
Common questions
- Does this change my existing Cursor rules?
- No. Setup merges a Concord block into your config and keeps existing rules and MCP servers. It is safe to re-run.
Further reading
Give your agents shared work-state.
npm install -g @concord-ai/concord-mcp