review

Review the current code diff using deterministic detectors and LLM reviewers.

2|1|Updated May 2, 2026
One-click install
npx skills add https://github.com/iroha924/mumei --skill review-iroha924
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/iroha924/mumei/tree/main/skills/review
Command: npx skills add https://github.com/iroha924/mumei --skill review-iroha924

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires Read, Grep, Glob, Bash, Task, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides a quick, one-shot review of code changes, ensuring that the code meets the highest standards without the full mumei spec/plan workflow.

Core Features & Use Cases

  • Code Review: Offers a one-shot, strongest-available code review of the current diff.
  • Flexible Use: Suitable for users who drive their own spec or are making small changes.
  • No Side Effects: Does not create .mumei state, write the ledger, or commit changes.
  • Use Cases: Triggered by commands like "/mumei:review", "review my diff", or "review this PR locally".

Quick Start

To review the current code changes, type: /mumei:review

Frequently Asked Questions about review

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

FAQPage Schema
How do I run a one-shot code review on my current diff?

To run a one-shot code review on your current diff, trigger the analysis using the /mumei:review command. This instantly evaluates your uncommitted code changes using deterministic detectors and diverse LLM reviewers without creating side effects.

Can I review a local PR diff without modifying my repository state?

Yes, you can review a local PR diff without side effects. The review process runs read-only analysis using Read, Grep, Glob, and Bash tools, ensuring it does not commit changes, write ledgers, or create persistent state files.

What is the difference between a one-shot LLM analysis and a full spec workflow?

A one-shot LLM analysis provides an instant review of current code changes without the overhead of a full spec or plan workflow. It is designed for developers making small changes or driving their own specs who need strong, immediate feedback.

Do I need specific tools enabled to perform an instant code review?

Yes, performing an instant code review requires the Read, Grep, Glob, Bash, and Task tools to be enabled. These dependencies allow the reviewer to scan files, search patterns, and execute deterministic detectors alongside LLM analysis.

When should I avoid using a one-shot diff review?

You should avoid a one-shot diff review if your task requires a comprehensive spec or plan workflow. This review is intended for quick, standalone analysis of uncommitted changes rather than managing extensive, multi-step development cycles.

How does deterministic code analysis work with LLM reviewers?

Deterministic code analysis works alongside diverse LLM reviewers by combining rule-based detectors with language model evaluation. This hybrid approach identifies potential issues in your diff instantly, providing a strong review without requiring a full spec workflow.