Skip to content

woodpecker-mcp

MCP server for Woodpecker CI. Stateless - each request carries its own authentication token, so the server scales horizontally without a database or shared secret.

The server exposes the Woodpecker API as MCP tools, so an AI assistant can query repositories, pipelines, logs, cron jobs, secrets, agents, organizations, users, and system info - and act on it.

The server isn't just a 1-to-1 mirror of the Woodpecker API - it ships AI-native tools that aggregate and analyze for you:

  • Diagnose pipeline failures - ask "Why did the last pipeline fail?" and get the failing step, its logs, and the config that ran, all in one call (explain_pipeline_failure).
  • Review CI config - ask "Is the CI config sound before we release?" and get deterministic lint findings on images, events, deprecated keys, and secret usage (review_pipeline_config).
  • Monitor pipelines - everyday questions about the queue, recent runs, logs, and scheduling, answered by the 50 thin read/action tools.

Features

  • Query repositories, pipelines, logs, cron jobs, secrets, agents, organizations, users, and system info
  • Trigger, restart, cancel, approve, and decline pipelines
  • Manage cron jobs and repository secrets
  • Diagnose pipeline failures with the AI-native explain_pipeline_failure tool
  • Lint CI configs with the deterministic review_pipeline_config tool
  • Per-request authentication (stateless, horizontally scalable)
  • Read-only MCP resources (woodpecker://) for repositories, pipelines, and the current user

Get started

Jump into Setup to install and run the server, head to the Usage Guide for the full list of tools, or start with a Featured workflow above.

Note

The server never stores your Woodpecker token. Each request must carry its own Authorization: Bearer <token> header, scoped to the permissions of that token.