Skip to content

Cursor

Cursor’s agent can run terminal commands and call MCP tools. Either works; the CLI keeps tokens out of the model’s context.

  1. Install the CLI and sign in:

    Terminal window
    curl -fsSL https://github.com/angler-ai/the-runbook-app/releases/latest/download/install.sh | sh
    runbook login https://runbooks.example.com

    Installing the CLI has the other ways, such as building from source.

  2. Install the skill:

    Terminal window
    runbook skill install --agent cursor --therunbook --scope project

    That writes .cursor/skills/therunbook/ in this repository. Without --scope project it goes to ~/.cursor/skills/.

  3. Give the agent its own credential:

    Terminal window
    runbook token create --agent --name cursor --expires-in 30d \
    --scope runbooks:read --scope executions:read --scope executions:act --json

    Export the secret as RUNBOOK_TOKEN in the terminal Cursor’s agent uses.

  4. Ask the agent, in agent mode:

    Work the next agent step of run exe_01JD2X6PZ5R8E2W4Y6A0C3D5F7.
.cursor/mcp.json
{
"mcpServers": {
"runbook": {
"command": "runbook",
"args": ["mcp", "serve"]
}
}
}

Put the file in ~/.cursor/mcp.json instead to make the server available in every project.

With the MCP server connected, Cursor’s agent can draft a runbook, validate it and save it as a draft (validate_runbook, save_runbook_draft). A person with runbooks:publish publishes it. See draft a runbook from an incident doc.