← スキル一覧に戻る
Thorough Mode (

code-review
by GGPrompts
⭐ 1🍴 0📅 2026年1月20日
SKILL.md
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:
- Main Review -
code-review:reviewer(Opus) - Security Scan -
code-review:security-scan(Haiku) - Silent Failures -
code-review:silent-failure-scan(Haiku)
4. Process Results
Collect output from all agents and:
- Merge findings - Deduplicate issues across agents
- Filter by confidence - Only report issues ≥80% confidence
- Check for blockers (security vulnerabilities, critical bugs, required tests missing)
- 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 |
スコア
総合スコア
50/100
リポジトリの品質指標に基づく評価
✓SKILL.md
SKILL.mdファイルが含まれている
+20
○LICENSE
ライセンスが設定されている
0/10
○説明文
100文字以上の説明がある
0/10
○人気
GitHub Stars 100以上
0/15
○最近の活動
3ヶ月以内に更新がある
0/10
○フォーク
10回以上フォークされている
0/5
✓Issue管理
オープンIssueが50未満
+5
✓言語
プログラミング言語が設定されている
+5
○タグ
1つ以上のタグが設定されている
0/5
レビュー
💬
レビュー機能は近日公開予定です