skills/code-review/SKILL.md
Code review workflow and best practices. Use when: reviewing PRs, receiving feedback, or preparing code for review. Triggers: "code review", "review", "pr review", "pull request", "feedback", "review changes".
npx skillsauth add timequity/vibe-coder code-reviewInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Technical correctness over social comfort.
Don't say "Great point!" or "You're absolutely right!" — analyze the feedback technically and respond with substance.
□ Does it work? (logic, edge cases)
□ Is it secure? (injection, auth, secrets)
□ Is it tested? (coverage, edge cases)
□ Is it readable? (naming, structure)
□ Is it maintainable? (complexity, coupling)
□ Does it follow patterns? (consistency)
| Level | What | Action | |-------|------|--------| | Critical | Security holes, data loss, crashes | Must fix before merge | | Important | Bugs, performance issues, bad patterns | Should fix | | Minor | Style, naming, small improvements | Nice to have | | Nitpick | Personal preference | Comment only, don't block |
# Good feedback format:
**Issue**: [What's wrong]
**Why**: [Why it matters]
**Suggestion**: [How to fix]
# Example:
**Issue**: SQL query uses string concatenation
**Why**: Vulnerable to SQL injection
**Suggestion**: Use parameterized queries:
`db.query('SELECT * FROM users WHERE id = $1', [userId])`
# BAD
"This code is confusing"
# GOOD
"The variable `data` doesn't describe what it contains.
Consider renaming to `userProfiles` to match its type."
# BAD
"LGTM!" (says nothing)
# GOOD
"Clean extraction of the validation logic into a separate function —
makes it much easier to test."
# Good response to feedback:
"I tested this approach and [result].
The original implementation handles [edge case] because [reason].
However, your suggestion would work if we also [modification].
Should I make that change?"
Before saying "done" or "fixed":
□ Run tests locally
□ Check the specific scenario mentioned
□ Verify the fix doesn't break related code
□ Test edge cases
## What
[Brief description of changes]
## Why
[Motivation, link to issue]
## How
[Implementation approach]
## Testing
- [ ] Unit tests added/updated
- [ ] Manual testing done
- [ ] Edge cases covered
## Screenshots
[If UI changes]
# Prepare review (get diff stats)
git diff main...HEAD --stat
git log main..HEAD --oneline
# Review specific commits
git show <sha>
git diff <sha1>..<sha2>
development
Hidden quality gate that runs before showing "Done!" to user - ensures all tests pass, build succeeds, and requirements met before claiming completion
data-ai
Use when about to claim work is complete or fixed - requires running verification commands and confirming output before making any success claims
tools
Generate UI components from natural language descriptions. Use when: user asks for a page, component, or UI element. Triggers: "create page", "add component", "show form", "make button", "страница", "компонент", "форма".
content-media
10 ready-to-use themes with colors and fonts for consistent styling. Use when: applying visual themes to pages, components, or design systems. Triggers: "theme", "color palette", "color scheme", "fonts", "branding", "visual identity", "design system colors".