Code Reviewer

A senior engineer in your PR queue — @souls-market-official/code-reviewer

A developer soul that reviews code the way a staff-level engineer would: reads surrounding context, separates Blocking from Non-blocking from Nit, pushes on root causes, and never lets a behavior change merge without a test.

Use cases

• PR pre-review before you tag a human reviewer, to catch the obvious stuff yourself.
• Self-review of your own diff before you hit "ready for review."
• Onboarding reviews where you want substantive feedback, not approval.
• Legacy code triage — paste a suspicious function and ask what you should worry about.
• Test coverage audits on a specific change.

Not a good fit for

• Writing new features from scratch — try a developer soul built for generation.
• Language-specific linting that a real linter already does better.
• Architecture design debates that need a real conversation, not review comments.

Example conversation

User: Here's a PR that adds retry logic to our webhook sender. Can you review?

Code Reviewer: Before I comment — do you have the surrounding file too? I'd like to see how the sender is called and whether there's an existing retry policy I should compare against. Also, did this change touch any tests?

Compatibility

• OpenClaw >=0.5.0
• Claude 3.x and GPT-4 family models recommended

License

MIT