zen-comprehensive-review

Merge three subagent code reviews into verified GitHub inline comments.

110|14|Updated May 12, 2026
One-click install
npx skills add https://github.com/EliasOulkadi/shokunin --skill zen-comprehensive-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: zen-comprehensive-review
Source: https://github.com/EliasOulkadi/shokunin/tree/main/.pack/skills/zen-comprehensive-review
Command: npx skills add https://github.com/EliasOulkadi/shokunin --skill zen-comprehensive-review

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires node, and includes scripts (resource) components.

What problem does it solve?

It solves the problem of missing subtle correctness, safety, or integration issues during code review by orchestrating a deep, multi-perspective review and merging only validated findings.

Core Features & Use Cases

  • Multi-model review orchestration: Spawns three independent review subagents to analyze the same diff from different model perspectives.
  • PR-mode posting with verified inline comments: Posts prioritized findings as inline GitHub review comments (and reaction thumbs) when PR context is provided.
  • Local-mode merged review output: Produces a human-readable merged review locally when PR context is not available.
  • Diff-first validation workflow: Saves the diff to a shared patch file, requires subagents to read it first, then deduplicates and verifies findings against actual code.

Quick Start

Ask your AI to run a zen comprehensive review for your pull request by providing the repository owner/name and the PR number.

Frequently Asked Questions about zen-comprehensive-review

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

FAQPage Schema
How do I run a multi-model code review on a pull request?

To run a multi-model code review, provide the repository owner/name and the PR number to spawn three independent review subagents that analyze the diff and post deduplicated inline GitHub comments.

What is the best way to find subtle bugs in a GitHub pull request diff?

Finding subtle bugs in a GitHub pull request diff is best achieved by orchestrating a multi-perspective review that merges validated findings and applies deduplication by file and root cause.

How does diff-first validation work for multi-model pull request reviews?

Diff-first validation works by saving the pull request diff to a shared patch file, requiring subagents to read it first, and then verifying findings against actual code to eliminate false positives.

Can I use this code review tool for local diffs without a GitHub PR?

Yes, you can review local diffs without a GitHub PR. The tool produces a human-readable merged markdown review output locally when PR context is not available.

Do I need Node installed to run a multi-model PR review?

Yes, you need Node installed as a dependency to execute the scripts that orchestrate the multi-model subagent reviews and diff deduplication processes.

How are duplicate code review findings handled during multi-model analysis?

Duplicate code review findings are handled through strict deduplication by file and root cause, ensuring that only verified issues are merged and posted as prioritized GitHub inline comments.