Day 30

AI-Powered Code Reviews

Use Claude for thorough, consistent code reviews — catching what humans miss.

0 / 6 tasks

Focused code reviews

bash

# Full review
git diff main...feature/my-branch | claude -p "Review this PR for: logic errors, edge cases, missing error handling, security issues, and test coverage. Be specific."

# Security-focused
git diff main...feature/auth | claude -p "Security review only. Look for: SQL injection, XSS, broken auth, exposed secrets, insecure deserialization."

# Performance-focused
git diff main...feature/bulk-import | claude -p "Performance review. Focus on: N+1 queries, unnecessary loops, missing indexes, memory usage."

Practice Tasks

0 / 6

Full PR review medium

Have Claude review your next PR before it goes to a human reviewer. Implement at least 2 findings.

Security review hard

Run a dedicated security review on a recent PR or change. Fix any vulnerabilities found.

Performance review medium

Run a performance-focused review on a database-touching change.

Test coverage review medium

Have Claude identify untested code paths in a recent change and generate the missing tests.

Compare AI vs human review hard

Have Claude review the same PR as a colleague. Compare findings. What did each miss?

Build a review script hard

Create a shell script that runs Claude's review process and formats the output as a markdown report.

← Day 29: Git + Claude Basics ↑ Module overview Day 31: Branch Strategy & Flow →