yeameen

yeameen / claude-code-review-council

Public

Multi-agent code review for Claude Code: parallel review by Codex (GPT-5.5), Gemini 3.1 Pro, and 4 Claude specialist subagents (security, performance, logic, regression)

ai-code-review claude-code claude-plugin code-review code-review-tool
11
1
85% credibility
Found May 23, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

This is a plugin for Claude Code that acts like a panel of expert code reviewers. When you have code changes to review, it launches six different AI reviewers simultaneously—one focused on security, one on performance, one on logic errors, one on regression risks, plus two more from different AI systems. Each reviewer looks at the same changes from their own angle. The system then combines all their findings, removes duplicates, verifies each issue against your actual code (to catch mistakes), prioritizes them by severity, and surfaces any disagreements between reviewers. You see a clean report and approve any fixes before they are applied. It's designed for important changes like security-sensitive code, database migrations, or anything where you'd want a senior engineer's eyes on it before merging.

How It Works

1
💻 You're working on important code

You have some changes you're about to merge—maybe a security fix, a database update, or a tricky feature—and you want extra eyes on it before it goes live.

2
🧩 You add the review council to your assistant

You install a special review tool that brings in multiple AI reviewers to check your work from different angles.

3
🔍 You ask for a thorough review

You simply ask your assistant to review your changes—whether it's uncommitted work, a specific commit, or even a GitHub pull request.

4
Six reviewers examine your code at once
🔐
Security specialist

Hunts for vulnerabilities and access control issues

⚡
Performance specialist

Spots slow queries and inefficient patterns

🧠
Logic specialist

Checks if the code actually does what it should

🔄
Regression specialist

Looks for things that might break elsewhere

🌐
Two AI reviewers

Additional perspectives from different AI families

5
✅ Every finding gets verified

The system checks each issue against your actual code to make sure it's real—not a mistake—and re-ranks the severity based on what's actually there.

6
📋 You see a clear, unified report

You get a prioritized list of findings (critical issues first, minor suggestions last), with disagreements between reviewers surfaced so you can make the final call.

🎉 Your code improves with your approval

You review the findings, approve the fixes you want, and your assistant applies them—running tests to make sure everything still works.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is claude-code-review-council?

This is a Claude Code plugin that runs six AI reviewers in parallel on your code changes. It combines Codex (GPT-5.5), Gemini 3.1 Pro, and four Claude specialist agents focused on security, performance, logic, and regression testing. The key innovation is the synthesis step that deduplicates findings, verifies citations against actual code to catch hallucinations, and surfaces disagreements between reviewers. You get a unified P0-P3 severity report before any code changes are applied.

Why is it gaining traction?

The multi-model approach is the hook - different AI families miss different issues, so running them all increases coverage. The verification step addresses a real pain point with AI code review tools: hallucinations. The four specialist subagents also serve as a fallback if external APIs fail. It integrates directly into Claude Code workflow with simple commands like `/review-council` or `/review-council 1234` for PR reviews.

Who should use this?

Security-sensitive code, database migrations, and schema changes benefit most from this approach. Teams wanting thorough pre-merge reviews will find it valuable, though individual developers working on smaller changes may find the 3-8 minute runtime excessive for simple fixes.

Verdict

This is a promising early-stage tool with solid architecture, though the 11-star count signals limited real-world testing. The MIT license and 0.85 credibility score suggest it's worth exploring for larger codebases where comprehensive review matters more than speed.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.