GoCD

You can connect your AI assistants to GoCD via a focused MCP server that exposes GoCD data and actions over the MCP protocol. This bridge lets you query pipelines, stages, and jobs, trigger runs, inspect failures, and retrieve logs and artifacts from GoCD in a structured, AI-friendly way.

How to use

To use the GoCD MCP Server, you connect your MCP client to the server’s endpoint and authenticate with a GoCD API token. The server exposes a standard MCP route at /mcp and accepts your token as a Bearer token. You can perform common operations such as listing pipelines, checking statuses, triggering pipelines or stages, reading logs, and analyzing failures. You can also paste natural language requests like “Show me why the build failed” or “Trigger the deployment stage,” and the MCP client will translate these into the appropriate GoCD actions.

How to install

Prerequisites: ensure you have Node.js installed (Node.js >= 18.0.0 is recommended). If you plan to run with Docker, you will need Docker installed as well.

# Install dependencies and run in development mode
npm install
npm run dev

# Optional Docker usage with the provided Dockerfile
docker build -t gocd-mcp .
docker run -d --env-file=.env -p 3000:3000 gocd-mcp

Configuration and environment

Configure the server by setting environment variables. The GoCD server URL is provided by GOCD_SERVER_URL. The MCP server binds to a host and port defined by MCP_HOST and MCP_PORT. Create a .env file in the project root with these values.

GOCD_SERVER_URL=https://your-gocd-server.com
MCP_HOST=0.0.0.0
MCP_PORT=3000

Available tools

list_pipelines

Browse all pipelines with their groups and pause status.

get_pipeline_status

Check if a pipeline is running, paused, or locked.

get_pipeline_history

View recent pipeline runs and their results.

get_pipeline_instance

Get detailed information about a specific pipeline run.

trigger_pipeline

Manually start a pipeline run.

pause_pipeline

Pause pipeline scheduling.

unpause_pipeline

Unpause pipeline scheduling.

get_stage_instance

View stage details including all jobs and their status.

trigger_stage

Manually trigger or retry a stage.

cancel_stage

Stop a running stage.

parse_gocd_url

Extract pipeline/stage/job info from GoCD URLs.

analyze_job_failures

Get comprehensive failure analysis including test results and console logs.

get_job_history

View job execution history.

get_job_instance

Check job status and details.

get_job_console

Read build logs and error output.

list_job_artifacts

Browse artifacts produced by a job.

get_job_artifact

Download specific artifact files.

parse_junit_xml

Extract structured test results from JUnit XML reports.