reviewer

Launch sub-agents to review local diffs and output per-role reports.

8|Updated Nov 4, 2016
One-click install
npx skills add https://github.com/coil398/dotfiles --skill reviewer-coil398
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reviewer
Source: https://github.com/coil398/dotfiles/tree/main/.claude/skills/reviewer
Command: npx skills add https://github.com/coil398/dotfiles --skill reviewer-coil398

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

reviewerエージェントにローカルの差分をレビューさせることで、バグ・セキュリティ・パフォーマンス・保守性・命名一貫性・リグレッション・データアクセス重複などの観点から評価を行い VERDICT: PASS/FAIL を返します。PR番号・リモートブランチ・gh pr 経由のレビューは /review-pr を使うこと。ユーザーが /reviewer と入力したら必ずこのスキルを使う。

Core Features & Use Cases

  • reviewer サブエージェントをハイブリッド並列で起動し、correctness/consistency/quality/security/architecture の観点で差分を総合評価します。
  • REVIEWER_SET の決定は、ユーザーフラグからの観点指定または自動選定に基づきます。
  • 各観点ごとに RUN_DIR に {RUN_DIR}/review-01-{ROLE}.md のレポートを出力し、全体の VERDICT を提示します。実務ケースとしては PR の差分、ブランチ差分、未コミット差分のレビューが含まれます。

Quick Start

Ask the reviewer skill to analyze the current diffs and generate per-role review reports under the RUN_DIR.

Frequently Asked Questions about reviewer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate code review for local diffs across multiple dimensions?

Automated code review for local diffs orchestrates multiple sub-agents to evaluate changes across correctness, consistency, quality, security, and architecture, outputting per-role reports and a final PASS or FAIL verdict.

Can I run an automated security and architecture check on uncommitted changes?

You can run automated security and architecture checks on uncommitted changes by applying multi-view code review to branch diffs, PR diffs, and local working tree modifications.

How does multi-agent code review work for evaluating branch diffs?

Multi-agent code review works by launching parallel sub-agents that independently assess diff correctness, consistency, quality, security, and architecture, generating isolated role-based reports under a designated run directory before producing a final verdict.

What is the best way to review code quality and consistency before creating a pull request?

The best way to review code quality and consistency before a pull request is using an automated multi-view reviewer that evaluates local diffs and outputs a comprehensive PASS or FAIL verdict with detailed role-based reports.

Does automated code review require Claude Code and a defined reviewer set?

Automated code review requires Claude Code and a defined REVIEWER_SET to determine evaluation perspectives, basing the selection on user-specified flags or automatic role assignment.

When should I use local diff review instead of reviewing a remote PR branch?

You should use local diff review for uncommitted changes and branch comparisons, while remote PR branches accessed through gh pr require a dedicated PR review process.