AI CodeOps for GitLab
Run AI CodeOps on GitLab.com or a self-managed GitLab: agents write the change on a branch, open a merge request, and it merges only once your gates pass.
AI CodeOps works with GitLab.com and self-managed GitLab instances. You connect GitLab once, then describe a change in the browser. Agents plan it, write it, review it and run the project's tests, and the result arrives in GitLab as a merge request that has already passed your quality gates.
What the integration does
- Repositories: Develop → Repositories lists the repositories the connected account can see, with their branches, files and merge requests.
- Projects: a project points at one repository. Sessions in that project clone it, write the change and test it in a sandbox. See projects.
- Branches and merge requests: when a change is approved and passes the release gate, AI CodeOps pushes a branch named
codeops/run-<id>and opens a merge request into the project's default branch. Commits are authored as "AI CodeOps". - Merging: a merge request the console opened can be merged from the console, and the release gate runs again first. You can also merge it in GitLab as usual.
AI CodeOps doesn't install webhooks, post commit statuses or comment on your merge requests. It never pushes to your default branch directly.
How to connect
Connect with GitLab OAuth, or paste a personal access token. OAuth asks for the api and read_user scopes and works against GitLab.com. A token needs the api scope; a project or group access token also needs the Developer role or higher. For a self-managed instance, enter its public https address; AI CodeOps runs in the cloud, so it must be able to reach the server over the internet.
The full steps, scopes and error messages are in Connect GitLab. Tokens are encrypted at rest (AES-256-GCM) and never written into a repository's Git configuration. To remove access, disconnect in AI CodeOps or revoke the token in GitLab.
Why teams use it with GitLab
Teams on GitLab often have no AI review built into their merge request flow at the tier they run. AI CodeOps adds review, tests and approvals before the merge request is opened, so reviewers see a change that has already passed the gate.
Every change carries its history with it: which model wrote it, what the reviewers found, which tests ran and who approved it. That history is kept as a signed evidence pack you can hand to an auditor. Spend is visible per session and capped per workspace, so an agent can't run up a bill unseen. See cost controls.
Getting started
- Create a free account: AI CodeOps is in open beta and runs in your browser.
- Connect GitLab from Configure → Settings → Git Providers.
- Create a project from a repository and run a first session. The quickstart walks through it.
AI CodeOps also works with the other Git hosts, so a team that uses more than one can run them all from one workspace.
For partners
AI CodeOps is a browser-based platform where AI agents plan, write, review and test code changes and open them as merge requests on GitLab. Changes merge only after tests, an AI review and the required approvals pass, and each one keeps a signed record of how it was made. Free during its open beta: https://aicodeops.dev/integrations/gitlab
Frequently asked questions
Does AI CodeOps need admin access to our GitLab?
No. It needs an account that can read the repositories you want to use, push a branch and open merge requests. The connection belongs to the person who made it.
Does AI CodeOps support self-managed GitLab?
Yes. Enter your instance's public https address when you connect. Servers on private networks can't be reached from the cloud service.
Does AI CodeOps open merge requests or push to main?
It opens a merge request from a new branch into the project's default branch. It never pushes to the default branch directly.
Can a merge request merge before it passes the gate?
Not through AI CodeOps. A merge request it opened can be merged from the console only after the release gate passes again: tests, AI review findings and approvals.