Connecting to the CircleCI CLI MCP Preview
| The CircleCI CLI MCP server is currently in preview. You can expect changes to the available tools during this phase. If you have any issues, questions, or feedback, post on our community forum. |
The CircleCI CLI includes a built-in MCP server that runs locally on your machine, using your existing CLI authentication. For a comparison with CircleCI’s hosted MCP server, see the Hosted and CLI MCP Comparison section of the CircleCI MCP overview page.
Prerequisites
-
The CircleCI CLI installed. See the CircleCI CLI page for installation instructions.
-
An authenticated CLI session. Run
circleci auth login, or set theCIRCLE_TOKENenvironment variable.
Enable the CLI MCP
Enable the MCP server for your AI assistant with one command:
circleci mcp claude enable # Claude Desktop
circleci mcp cursor enable # Cursor
circleci mcp vscode enable # VS Code
To disable it, replace enable with disable.
You can also start the server manually, or stream it over HTTP instead of the default stdio transport:
circleci mcp start # start over stdio
circleci mcp stream # stream over HTTP
Try it out
Once enabled, ask your AI assistant something like:
-
What failed on my last run? -
Validate my CircleCI config. -
List the environment variables set on this project. -
Create and publish a new orb version. -
Show me the resource classes for my self-hosted runners. -
Purge the Docker Layer Cache for this project.
Next steps
-
CircleCI MCP Overview, for how the CLI MCP compares to the hosted server, best practices, and troubleshooting.
-
Connecting to the Hosted CircleCI MCP, for the hosted alternative.
-
CircleCI CLI reference: MCP, for the full command reference.