review

Automate multi-perspective pull request and merge request reviews by URL or local diff.

1|Updated Mar 3, 2026
One-click install
npx skills add https://github.com/AndreJorgeLopes/devflow --skill review-andrejorgelopes
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/AndreJorgeLopes/devflow/tree/main/skills/review
Command: npx skills add https://github.com/AndreJorgeLopes/devflow --skill review-andrejorgelopes

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents slow, error-prone PR/MR reviews by automatically gathering deep context and running multiple specialized review perspectives in parallel so issues are caught early and consistently.

Core Features & Use Cases

  • URL or local diff review: Reviews a GitHub PR/MR by URL, a GitLab MR by URL, or local staged/uncommitted changes against CLAUDE.md.
  • Thorough vs quick vs spec modes: Adapts review rigor based on diff size and whether the changes are primarily markdown/spec/skill files.
  • Multi-agent findings with scoring: Runs role-based agents (bug scanning, CLAUDE.md compliance, test coverage, plan alignment, git history, sibling MR/epic coherence) and outputs confidence-scored, structured findings.

Quick Start

Ask your agent to run /devflow:review on your PR URL in thorough mode.

Frequently Asked Questions about review

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

FAQPage Schema
How do I automate pull request reviews with multiple agents?

Automated multi-perspective pull request reviews run specialized agents in parallel to scan bugs, check test coverage, and verify compliance. This approach dispatches role-based agents with a shared context packet to emit deduplicated, confidence-scored findings.

Can I review local git diffs or staged changes without a URL?

Local git diffs and staged changes can be reviewed directly against CLAUDE.md specifications. The review process applies to local uncommitted changes by fetching the diff context and dispatching agents to analyze the modifications.

What is the best way to review markdown or spec files in a merge request?

Markdown and spec file reviews adapt automatically based on diff composition using a dedicated spec mode. This mode shifts focus to prose reviews, adjusting the review rigor when changes are primarily markdown, spec, or skill files.

Does this code review approach check existing PR discussions?

Existing PR discussions are checked by cross-referencing new findings against current conversation threads. Mandatory cross-checking against existing discussions ensures the output only emits deduplicated, confidence-scored findings.

How do I run a quick code review instead of a thorough analysis?

Quick code reviews are executed by selecting a quick single-pass mode instead of thorough mode. Mode arguments adapt the review rigor based on diff size, allowing faster feedback for smaller pull requests.

What perspectives are covered during a multi-agent code review?

Multi-agent code review perspectives include bug scanning, CLAUDE.md compliance, test coverage analysis, plan alignment, git history, and sibling MR coherence. Role-based agents run in parallel to provide comprehensive coverage.