Maintainers carry a lot of the software world. If AI agents are going to touch their projects, they should do it behind the same quality gates as everyone else, at no cost.
Open-source maintainers are in an odd spot right now. AI tools make it easy for anyone to send a pull request, so the volume of contributions goes up. The time a maintainer has to review them does not.
The usual answers are to review faster or to say no more often. Neither scales. What does scale is making every change prove itself before a person spends time on it.
That's the idea behind AI CodeOps, and it's why we're making it free forever for non-commercial open-source projects.
What maintainers get
The open source plan isn't a trial. Approved projects get, at no cost:
- Higher limits than the free plan: more workspaces, seats and AI models
- 300 sessions a month on the platform's models. Past that, you can keep going on your own model provider key, or wait for the monthly reset
- The release-gate badge for your README, turned on for you
- A usage meter in the console, so you always know where you stand
It's the same product, with the same gates. Nothing is held back for paying customers when it comes to what a change must pass before it merges.
What the gate does for a project
Every AI change in AI CodeOps waits at a release gate. It becomes a pull request only when:
- Your own test suite passed in a sandbox, on that exact version of the change
- The AI review of the real diff has no open high or critical findings
- A person approved it. The person who started the session can't cast the approving vote while someone else in the workspace can
If any check errors, the gate fails closed. For a maintainer, that means the pull request in your queue has already been tested and reviewed, and someone has looked at it.
Each change also carries a signed evidence pack: the task, the models used, the findings, the test runs and the approvals. When someone asks how a change got in, there's a record.
Your Git host, your server
A lot of open source doesn't live on GitHub. AI CodeOps works with GitHub, GitLab, Bitbucket, Gitea and Forgejo, including self-hosted GitLab, Gitea and Forgejo servers connected with a personal access token. Branches and pull requests stay on your server.
One honest note: if your project is on Codeberg, read its Terms of Use first. Codeberg doesn't allow projects that mostly consist of code written by generative AI tools. The connection works, but the rules are Codeberg's to set.
Who qualifies
- A public repository under an OSI-approved license
- A non-commercial project: not sold, not the core of a paid product, and not run by a company to make money from it. Community projects, libraries and tools qualify, even with sponsors or donations
- Approved by application. We read every one, usually within a few working days, and email you either way
How to apply
- Sign up at aicodeops.dev with the email you'll use for the project. Approval moves that account's own organization onto the open source plan
- Fill in the short form at aicodeops.dev/pricing/open-source: email, repository URL, maintainer's GitHub username, license, and a few words about the project
And as with every plan: AI CodeOps doesn't use your code, prompts or outputs to train AI models.
If you maintain something people depend on, we'd like to help you ship changes you can trust. Apply here: aicodeops.dev/pricing/open-source. Program updates are posted at github.com/aicodeops-dev/announcements.
Drafted with AI assistance and edited by the AI CodeOps team.
Top comments (0)