Install guide and config at curatedmcp.com

GitLab MCP: Automate Your Entire DevOps Workflow Inside Claude

If you're using GitLab, you already know your whole development lifecycle lives there—issues, merge requests, pipelines, deployments. The GitLab MCP server is the official bridge that lets Claude and other AI agents access all of it, so you can automate code review, incident triage, and release coordination without context-switching.

What It Does

GitLab MCP gives AI agents real, executable access to your GitLab workspace. Instead of manually describing a problem to Claude and then running commands yourself, you can ask Claude to search issues, inspect pipeline runs, review code diffs, and even trigger deployments—all in one conversation.

The server unlocks:

• Issue and MR management — search, filter, and create issues; review merge request diffs and comments
• CI/CD automation — trigger pipelines, inspect logs, and check deployment status
• Release coordination — manage milestones, track progress, and automate alert response
• Project insights — pull metrics and activity reports to understand team velocity or incident patterns

It's designed for teams that actually use GitLab's full feature set—not just as a repo host, but as a DevOps platform. That means it works with epics, boards, deployments, and incident workflows, not just basic git operations.

How to Install

Install via npm:

npx -y @gitlab/mcp

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "gitlab-mcp": {
      "command": "npx -y @gitlab/mcp",
      "env": {
        "GITLAB_TOKEN": "your-personal-access-token",
        "GITLAB_BASE_URL": "https://gitlab.com"
      }
    }
  }
}

You'll need a GitLab personal access token with read_api and write_repository scopes. Full setup docs.

Real-World Use Cases

• Code review automation — Ask Claude to review all open MRs for a given project, flag risky changes, and summarize findings in a single report.
• Incident triage at scale — When alerts fire, have Claude search related issues, check recent pipeline failures, and draft a summary with linked deployments—no manual digging.
• Release checkpoints — Before deploying to production, prompt Claude to verify milestone completion, check pipeline status, and confirm all blockers are resolved.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.