Does it show the raw output from both models?

To keep your workspace clean, it suppresses raw intermediate outputs and only presents a synthesized final report that merges the best insights from both reviewers.

Can I use this for documentation or markdown files?

No, this skill is specifically designed for code. It skips documentation and trivial single-line changes to focus on logic, security, and code quality.

What happens if one of the review models fails?

The skill is designed to prioritize the 'council' aspect; if a reviewer fails, it will report the error rather than presenting a single-model result as a unified council report.

What are the prerequisites for using this skill?

You need the Codex CLI installed globally via npm and an authenticated session, in addition to the Claude Code environment.

How does Council Review differ from a standard AI review?

While standard tools use one model, Council Review runs two independent models (Codex and Claude) in parallel and cross-validates their results to minimize noise and maximize accuracy.

Council Code Review

Name: Council Code Review
Author: trancong12102

bytrancong12102

•

Collaboration Tools

Orchestrates a multi-model code review process by running Codex and Claude in parallel to deliver a single, cross-validated audit report.

Council Code Review acts as a virtual board of experts for your codebase, executing independent reviews via Codex and Claude’s internal tools simultaneously. By analyzing changes in uncommitted code, specific commits, or branch diffs, it cross-references findings to eliminate false positives and enhance the depth of feedback. Instead of overwhelming the user with raw logs, it synthesizes the collective intelligence into a single, prioritized, and deduplicated report, ensuring that only high-confidence, actionable improvements reach the developer.

Key Features

01Parallel multi-model analysis using Codex and Claude for diverse perspectives.

02Automated synthesis of findings into a single, unified markdown report.

033 GitHub stars

04Support for uncommitted changes, branch diffs, and specific commit SHAs.

05Cross-validation logic to filter false positives and enhance valid findings.

06Intelligent priority sorting from P0 to P4 to focus on critical issues first.

Use Cases

01Deep-cleaning legacy commits by cross-referencing multiple AI analysis models.

02Auditing a branch diff to prepare a comprehensive summary for a Pull Request.

03Performing a pre-commit sanity check to catch regressions before they are pushed.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add trancong12102/agentskills council-review

For use in Claude.ai and ChatGPT

Key Features

01Parallel multi-model analysis using Codex and Claude for diverse perspectives.

02Automated synthesis of findings into a single, unified markdown report.

033 GitHub stars

04Support for uncommitted changes, branch diffs, and specific commit SHAs.

05Cross-validation logic to filter false positives and enhance valid findings.

06Intelligent priority sorting from P0 to P4 to focus on critical issues first.

Use Cases

01Deep-cleaning legacy commits by cross-referencing multiple AI analysis models.

02Auditing a branch diff to prepare a comprehensive summary for a Pull Request.

03Performing a pre-commit sanity check to catch regressions before they are pushed.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add trancong12102/agentskills council-review

For use in Claude.ai and ChatGPT