code-review | Skill Performance & Reviews | TopRankSkills

TopRank Skills

Home / Skills / tools / code-review

code-review

maintained by GGPrompts

star 1 account_tree 0 verified_user MIT License
bolt View GitHub

name: code-review description: "Review code changes with automated quality checks and confidence-based filtering" user-invocable: true

Review code changes for beads issues with automated quality checks, confidence-based filtering, and parallel specialized agents.

Usage

/code-review                    # Review uncommitted changes
/code-review <issue-id>         # Review changes for specific beads issue
/code-review --thorough         # Deep review with parallel specialists

Process

Follow these steps precisely:

1. Determine Scope

If <issue-id> provided:

  • Get issue details: bd view <issue-id>
  • Get commits for issue: bd log <issue-id> or check issue metadata for commit range
  • Get diff: git diff <base-sha>..<head-sha>

If no issue-id:

  • Review uncommitted changes: git diff HEAD
  • Get list of changed files: git status --short

2. Find Relevant CLAUDE.md Files

# Get directories with changes
CHANGED_DIRS=$(git diff --name-only | xargs -I{} dirname {} | sort -u)

# Read root CLAUDE.md
cat CLAUDE.md 2>/dev/null

# Read CLAUDE.md in changed directories
for dir in $CHANGED_DIRS; do
  cat "$dir/CLAUDE.md" 2>/dev/null
done

3. Launch Review Agents

Standard Mode (default)

Spawn main reviewer agent:

Task(
  subagent_type="code-review:reviewer",
  prompt="Review changes. Issue: <issue-id>. Changed files: <files>"
)

The reviewer will:

  • Check CLAUDE.md compliance
  • Scan for bugs (confidence ≥80)
  • Auto-fix high-confidence issues (≥95)
  • Assess test coverage needs
  • Return structured JSON output

Thorough Mode (--thorough)

Launch 3 parallel specialized reviewers:

  1. Main Review - code-review:reviewer (Opus)
  2. Security Scan - code-review:security-scan (Haiku)
  3. Silent Failures - code-review:silent-failure-scan (Haiku)

4. Process Results

Collect output from all agents and:

  1. Merge findings - Deduplicate issues across agents
  2. Filter by confidence - Only report issues ≥80% confidence
  3. Check for blockers (security vulnerabilities, critical bugs, required tests missing)
  4. Auto-fixes applied - List what was fixed automatically (≥95% confidence)

5. Report Results

If blockers found:

❌ Code Review FAILED

BLOCKERS (must fix):
- [SECURITY] Exposed API key in src/config.ts:12
- [CRITICAL] Null pointer access in src/api.ts:45

Fix these issues and run /code-review again.

If warnings only:

⚠️  Code Review PASSED with warnings

Auto-fixed (2):
- Removed unused import in src/utils.ts:5

Important issues (3):
- Missing error handling in src/api.ts:34

Review complete.

If clean:

✅ Code Review PASSED

- Reviewed 5 files
- CLAUDE.md compliance verified
- No security issues
- No blockers

Ready to proceed.

Confidence Scoring

Score Meaning Action
0-79 False positive / uncertain Skip
80-94 Verified issue Flag
95-100 Certain bug or violation Auto-fix

chat Comments (0)

chat_bubble_outline

No comments yet. Be the first to share your thoughts!

Skill Details

GitHub Stars 1
GitHub Forks 0
Created Jan 2026
Last Updated 5 months ago
tools tools automation tools

Related Skills

specs-gen
chevron_right
glm-coding-agent
chevron_right
feature-dev
chevron_right
creating-pr
chevron_right
writing-skills
chevron_right

Build your own?

Join 12,000+ developers contributing to the Claude ecosystem.