code-review

Coordinate parallel style, logic, and security reviews of open GitHub PRs.

50|4|Updated Mar 21, 2026
One-click install
npx skills add https://github.com/sliamh11/Deus --skill code-review-sliamh11
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/sliamh11/Deus/tree/main/.claude/skills/code-review
Command: npx skills add https://github.com/sliamh11/Deus --skill code-review-sliamh11

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill automates thorough PR reviews by running three specialized agents (style, logic, security) in parallel, delivering structured findings and a learning feedback loop to reduce false positives over time.

Core Features & Use Cases

  • Parallel, multi-agent review of the current branch's PR
  • Confidence-scored findings with negative examples fed back into evolution
  • Context gathering from PR diff, files, body, and CLAUDE.md sources

Quick Start

Ask the assistant to run code-review on the active PR branch to start the automated analysis.

Frequently Asked Questions about code-review

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

FAQPage Schema
How do I automate PR review for style, logic, and security on a git branch?

Automated PR review coordinates three specialized agents—style, logic, and security—that run in parallel on your active branch's PR to deliver confidence-scored findings. It gathers the diff, file stats, and PR description to guide the analysis.

What is multi-agent code review and how does it work with GitHub?

Multi-agent code review executes specialized reviewers for style, logic, and security simultaneously on an open GitHub PR. It loads CLAUDE.md files and review criteria to guide the process, then consolidates findings with confidence scoring.

Do I need an authenticated GitHub CLI session to run automated code review?

Yes, automated code review requires an authenticated GitHub CLI session and a valid git repository with an open PR on the current branch. These prerequisites allow the skill to gather PR diffs, file stats, and descriptions for analysis.

How can I reduce false positives in automated PR reviews?

Automated PR review reduces false positives by recording results to review-history.log and loading dismissed findings to guide future analysis. This learning feedback loop feeds negative examples back into the review evolution.

Does multi-agent code review respect existing review history and dismissed findings?

Yes, multi-agent code review loads dismissed findings and existing review history from resources/review-history.log before analyzing. This ensures the specialized agents respect previous decisions and reduce duplicate false positives.

Can I run parallel security and style checks on my current pull request?

Yes, you can run parallel security and style checks by executing the multi-agent code review on your active PR branch. The skill coordinates three specialized reviewers—in style, logic, and security—simultaneously to deliver consolidated, confidence-scored findings.