AI CodeOps

AI CodeOps for GitLab, Gitea and Forgejo teams

AI coding agents for teams on GitLab, Gitea or Forgejo: connect with OAuth or a token, then run sessions with tests, review and approvals before merge.

Many AI coding tools start with GitHub and add other hosts later, if at all. If your code lives on GitLab, Gitea or Forgejo, AI CodeOps treats your host as a first-class option: the same agents, workflows and gates you would get on GitHub.

The problem

Teams that chose GitLab, or run their own Gitea or Forgejo server, often did it for control: over where code lives, who can see it, and how it is reviewed. When they look at AI coding agents, they find tools that assume GitHub, or that support other hosts with fewer features. The choice becomes "move your code" or "do without", and neither is acceptable.

How AI CodeOps helps

Connect the way your host expects

  • GitLab: connect GitLab.com or a self-managed instance. See the GitLab setup guide.
  • Gitea: enter your server's URL and connect with a personal access token. See the Gitea guide.
  • Forgejo: the same, with your Forgejo server's URL and a token. See the Forgejo guide.

Tokens are checked against your server before they are saved, and they are stored encrypted. Your server needs a public HTTPS address, because the agents run on the AI CodeOps platform and reach it from there.

The same features on every host

Once connected, you create a project from a repository and run sessions exactly as a GitHub team would. Agents plan the task, write the change, run your tests and review the diff for correctness and security. The result arrives on your host as a change ready for review, and it waits at the release gate until tests pass, AI findings are reviewed and the required approvals are in.

Governance that does not depend on your host

Approvals, policies and the audit log live in AI CodeOps, so they work the same whichever host a repository is on. Each change has a signed evidence pack that records the task, models, approvals, findings, tests and the resulting change, and anyone can check it at /evidence/verify. That is useful when your Git server is private and an auditor has no access to it.

Mixed estates

Plenty of organisations have GitLab for one product, a Gitea server for internal tools and a few repositories on GitHub. One workspace can connect several hosts, so every team gets the same workflows, the same spend caps and the same gate. See the GitLab integration for details on the connection.

Costs and models

Sessions run on the platform's Economy, Standard or Premium tiers, or Auto, or on your own provider key. Spend caps apply per workspace and per workflow, and each session shows its cost.

What to try first

  1. Follow the quickstart to create a workspace.
  2. In the workspace's Git settings, connect GitLab, or enter your Gitea or Forgejo URL and a personal access token.
  3. Create a project from one repository.
  4. Start a session with a small task and the "Plan, Build, Review & Test" workflow.
  5. Review the result on your host and in the release gate, then approve it or ask for changes.

Good to know

AI CodeOps is browser-only and runs on our hosted platform. If your policy requires keeping everything inside your own network, self-hosting is available as an Enterprise option: contact sales. AI CodeOps is free during the open beta.

Frequently asked questions

Does AI CodeOps work with self-managed GitLab?

Yes. GitLab connects for GitLab.com or a self-managed instance. Give the instance URL when you connect.

How do we connect our own Gitea or Forgejo server?

Enter the server's URL and a personal access token in the workspace's Git settings. The token is checked before it is stored, and it is stored encrypted.

Does our Git server need to be reachable from the internet?

Yes. AI CodeOps runs on its own servers, so your Git server needs a public HTTPS address it can reach.

Can we keep AI CodeOps entirely inside our network?

Self-hosting is available only as an Enterprise option. Contact sales to discuss it.