Code Review Agent
Reads a GitHub repository and reviews architecture, correctness, security, and tests.
- Typical run
- ~8 min
- Tools
- GitHub API · Repository read
- You provide
- A repository · A branch · A focus
- Approval
- Read-only: nothing leaves the platform
◤ What it does
Reads a GitHub repository and reviews architecture, correctness, security, and tests.
Connect a GitHub repository and the Code Review Agent reads its real source, then returns a structured review with findings ranked by severity: architecture, correctness, security, and test coverage, each with a concrete fix. Hand the findings to the Fix Agent and they come back as verified changes in a pull request.
Needs: GitHub repository read.
Limits: Reads code; does not compile or run it.
◤ Ask it
In your words. No prompts to engineer.
“Review the overall architecture and how the modules fit together.”
“Hunt for bugs and security gaps in this service.”
“Grade code quality and release readiness, blockers first.”
◤ Missions it runs in
- Audit my software
◤ Questions
About the Code Review Agent
Does it need write access to my repository?
No. Review needs read access only. Write access is requested separately, and only if you turn on the Fix Agent.
Can it fix what it finds?
Yes. Press Fix these findings under a finished review and the Fix Agent edits the code in an isolated sandbox, runs your build and tests, and opens a pull request.
◤ Works alongside
Agents that share its missions.
QA Agent
Drives a real browser against your live app, and every finding carries evidence.
Fix Agent
Turns review and QA findings into verified edits and opens the pull request.
Analyst
Reasons over the evidence earlier steps gathered, as a named specialist, labelling every claim.
Evaluator
Re-fetches sources, recomputes figures, and scores every deliverable before you see it.