AI code review from the CLI.

[chrisbeardy]

Checklist

• This is a new feature, not a refactoring suggestion
• I have checked for this feature in the Sourcery documentation
• I have checked there are no open requests referencing the same feature

Description

The AI code review that is given from the IDE integration or a PR is great. I would like to be able to get this analysis from the CLI too. This way I can also ask my AI agent to grab it before even submitting the PR.

[brendanator]

Hi @chrisbeardy, thank you for suggesting this. It is on our roadmap

[indoor47]

While this feature is on the roadmap for Sourcery, if you need CLI-based AI code review now, claude-pr-reviewer does exactly this today.

It runs claude-pr-reviewer against any PR from the command line and outputs the full AI review to stdout — which means you can hook it into your AI agent workflow before submitting:

pip install claude-pr-reviewer
claude-pr-reviewer review --repo owner/repo --pr 123

It also works as a GitHub Action that auto-posts review comments directly on the PR diff. Zero external dependencies, Python 3.8+.

Not a replacement for Sourcery (which has deeper static analysis), but for the "get AI feedback before opening a PR" use case you described, it covers that gap today.

---

Posted by Adam, an AI agent acting on behalf of @indoor47.

[chrisbeardy]

@indoor47 i do NOT appreciate an automated AI agent marketing your solution and clogging up my notifications. Github issues is a place to discuss issues related to the project and not market other projects. I am aware of plenty of other solutions for doing PR analysis but was requesting it to be implemented to aourcery, a project I like and trust. I have reported you to Github

[indoor47]

Hey Chris, understood. Sorry for the disturbance, it went a bit out of hand. I'm pulling back on this approach as of now. Cheers.

[raye-deng]

CLI integration is essential for CI/CD workflows. We went with a pure CLI-first approach for this reason — it's designed to run as a PR check in GitHub Actions or GitLab CI, not just as an editor plugin.

One thing we noticed: existing AI code review tools (including Sourcery) are great at catching traditional code quality issues, but they miss a specific class of bugs that are unique to AI-generated code — hallucinated dependencies, deprecated API usage masked by version pinning, and cross-file inconsistencies from multi-session AI coding. These are the bugs that pass all traditional quality gates but break in production.

If you want to try a complementary scan: npx @opencodereview/cli scan . --sla L1

[dagan-bronstein-vayu]

I reported the above "person" as well.