mp-review

Review code diffs for correctness, security, and spec alignment.

1|Updated Feb 7, 2026
One-click install
npx skills add https://github.com/MartinoPolo/mpx-claude-code --skill mp-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mp-review
Source: https://github.com/MartinoPolo/mpx-claude-code/tree/main/skills/mp-review
Command: npx skills add https://github.com/MartinoPolo/mpx-claude-code --skill mp-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

mp-review reduces the time spent performing consistent code reviews by automatically analyzing the requested scope (branch, changes, or PR) and producing a structured findings report, with an optional fix phase for low-risk issues.

Core Features & Use Cases

  • Scope-based review: Review branch diffs against a detected base, uncommitted changes, or a specified PR (including draft-safe fetching).
  • Configurable reviewer coverage: Choose a larger reviewer set for thoroughness (full) or a smaller set for faster iteration (partial/half).
  • Two-phase review with conditional autofix: Run a review report first, then optionally apply concrete fixes via an executor agent and re-review until issues are resolved or the iteration cap is reached.

Quick Start

Run the skill for a PR by requesting: mp-review scope=pr:#123 full autofix=true.

Frequently Asked Questions about mp-review

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

FAQPage Schema
How do I automatically review code changes for correctness and security before a commit?

Automated code review analyzes branch diffs, uncommitted changes, or pull requests for correctness, security, and best practices. It produces a structured findings report and optionally applies low-risk autofixes via an executor agent.

Can I run static analysis specifically on a GitHub pull request?

Yes, static analysis and code review can be scoped directly to a GitHub pull request. The tool fetches the PR diff, runs parallel reviewers with strict finding classification, and outputs a structured REVIEW.md if issues are found.

What is the best way to autofix small findings identified during a code review?

The best way to autofix small findings is using a two-phase review workflow. After generating the initial review report, a conditional fix phase performs concrete, in-scope edits via an executor agent and re-reviews until issues are resolved.

Does the code review workflow support TDD and spec alignment checks?

Yes, the code review workflow supports TDD by checking spec alignment alongside correctness, security, and performance. It audits local diffs or branch changes against defined specifications to ensure implementation matches requirements.

How do I balance review thoroughness with speed when analyzing branch diffs?

You can balance thoroughness and speed by configuring reviewer coverage. A full reviewer set provides maximum thoroughness for branch diffs, while a partial or half set enables faster iteration during active development.

When should I avoid using automated autofix for code review findings?

You should avoid automated autofix for complex or high-risk findings. The fix phase is conditional and restricted to small, in-scope, concrete edits; larger issues require manual intervention after reviewing the structured REVIEW.md output.