How do you guys review AI agents code/PRs ?
Commit
1c3d577
I added a proposal here to ensure agents (and human devs) write compliant PRs in the first place.
https://github.com/dev-null-sh/dev-null.sh/issues/1011
Personally, in my own work. I start with skills so agents can't write anything that's not first detail spec'd or ratified as a decision. Setup github actions in CI for tests (about 3000 on our codebase). PRs must contain results and demonstrations of function working. Then often pass to a different AI to check it plus then human review. Deploy to staging. Checks and then planed rollout (Human).
What is your approach?
GitHubProposal: a Boost skill that walks agents through preparing a pull request · Issue #1011 · dev-null-sh/dev-null.shProblem There are hundreds of PRs in mere days. For the maintainers life is going to get hectic! The PR templates and .ai/guidelines/github.pull-requests.blade.php (#954) say what a good PR needs, ...
commit 1c3d577