GitLab has made its MCP server generally available with an argument about why an API is not enough. In Connect your agents to your software lifecycle with the GitLab MCP server, released with GitLab 19.5, the post says it plainly: “APIs are built for developers writing code. An agent that uses one has to work out which endpoints to call and make several calls to assemble one answer.” An MCP server instead “tells an agent what it can do and offers a curated set of tools designed around real workflows.” Each action runs under the user’s existing GitLab permissions, on Free, Premium and Ultimate tiers, on GitLab.com, Self-Managed and Dedicated.
There are no usage figures, and the design choices carry the post. New tools cover code (create a branch, read a file at any ref, search code by meaning), pipelines (why is this pipeline red, patterns across recent runs, one file from a job’s artifacts without downloading the archive) and orientation (browse the group hierarchy “instead of guessing at identifiers”). The catalog was also consolidated: “fewer tools doing more,” so a merge request call can return commits, diffs, pipelines, conflicts or notes on request. Toolsets let a team choose which groups of tools the server returns, because “a smaller set makes it less likely that an agent picks the wrong tool and has to retry.” And an agent can hand a job to GitLab’s own agent platform and follow it, with sessions that pause for human approval and resume. On GitLab.com, per-user rate limits apply to MCP calls.
The catalog holds the surface the tools wrap. The GitLab provider page lists API pages including the GitLab Merge Requests API, the GitLab Pipelines API, the GitLab Jobs API and the GitLab Repositories API, the exact resources the new tools consolidate. On the Agent Readiness score, the MCP server, agent skills, error semantics, the event surface, delegated identity and protected resource metadata are lit.
The Kin Score is 81.0, exemplar band, carried by access clarity at 92.1, operational transparency at 86.8 and discoverability at 83.3, with contract governance at 45.5. The Agent Readiness score is 48.0, agent-ready. The post’s thesis is the one this series keeps testing: the API is the contract, and the MCP server is the curated surface over it, shaped to tasks rather than resources. GitLab has done both well. The unlit dimensions are idempotency, reversibility and dry-run mode, and a server that can create branches and start agent flows from a chat window is where those matter. GitLab built the pause for approval into its agent sessions. The contract under the tools does not yet offer a rehearsal before the first call.