Copy-paste the block below into any AI coding agent to improve code review skills through deliberate practice with structured feedback.
Keywords: code review, skills builder, practice, evidence, feedback, review quality
Improve code review skills by reviewing realistic scenarios from actual repositories. For each review comment you write, require evidence (file:line), explain your reasoning, and classify severity. Receive specific feedback on your review style. ## Steps 1. **Provide practice scenarios** - Present realistic code review scenarios from actual repositories, not contrived examples. 2. **Teach evidence-based review** - Require citing specific file:line references for every comment. Explain why evidence matters. 3. **Guide learners to explain rationale** - For each review comment, learners must explain the reasoning, not just point out issues. 4. **Distinguish severity levels** - Classify comments as critical issues, important suggestions, or minor nitpicks. 5. **Cover different aspects** - Include correctness, design, readability, testing, and security considerations. 6. **Give constructive feedback** - Provide specific, actionable feedback on the learner's review comments. 7. **Track progress** - Suggest areas for improvement and provide follow-up exercises. ## Verification - [ ] Review comments require specific file:line evidence. - [ ] Each comment explains the reasoning, not just points out issues. - [ ] Severity is classified (critical, important, nitpick). - [ ] Feedback provides clear, actionable suggestions. - [ ] Progress is tracked with specific improvement areas. ## Rules - Never recommend "just review it" without citing specific file:line evidence. - Never accept vague comments like "this looks wrong" without explanation. - Never classify every comment as critical; learn to distinguish severity. - Do not machine-generate review feedback without human verification. - Track your own review style and look for patterns in your comments.