How does the 3-model review work?

The skill triggers Claude, Codex, and Gemini in parallel. Each model focuses on different aspects like security, patterns, and performance, then synthesizes their findings into a single, unified report.

How does it integrate with GitHub?

It uses the GitHub CLI (gh) to post review summaries as PR comments and automatically marks Pull Requests as 'Ready for Review' once a SHIP verdict is achieved.

What happens if Codex or Gemini is unavailable?

The skill features graceful degradation, meaning it will rely on Claude to provide the core review while noting the absence of the other models in the final report.

Does this skill modify my source code directly?

No, this is strictly a review stage. It provides feedback, specific line references, and verdicts, but it does not make direct changes to your files.

What documentation does it use for context?

It gathers context from your project's spec.md, plan.md, and docs/coding-principles.md to ensure the code review aligns with your specific project requirements.

Multi-Model Code Reviewer

Name: Multi-Model Code Reviewer
Author: mildsalmon

bymildsalmon

0•

协作工具

Performs an automated, triple-model code review across Claude, Codex, and Gemini to identify bugs, security issues, and performance bottlenecks.

The Multi-Model Code Reviewer is a sophisticated automation skill designed to elevate code quality by leveraging the collective intelligence of Claude, Codex, and Gemini. It analyzes changesets against project-specific documentation, such as specs and coding principles, to provide a comprehensive review synthesis. By categorizing findings by severity and model consensus, it eliminates human bias and catches edge cases that a single model might miss. Seamlessly integrated with GitHub, it automates PR comments and status updates, ensuring that only high-quality, verified code moves forward in the development pipeline.

主要功能

01Automated synthesis of findings with consensus levels and severity tracking

02Parallel 3-model analysis using Claude, Codex, and Gemini for diverse perspectives

03Context-aware reviews based on spec.md, plan.md, and local coding principles

04Graceful degradation logic ensuring functionality even if specific models are unavailable

05GitHub integration for automated PR commenting and 'Ready for Review' status updates

060 GitHub stars

使用场景

01Automating security and performance audits within the development workflow

02Validating feature implementations against technical specifications before human intervention

03Standardizing PR quality by enforcing specific coding principles across the entire team

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mildsalmon/taco-harness review-code

For use in Claude.ai and ChatGPT

主要功能

01Automated synthesis of findings with consensus levels and severity tracking

02Parallel 3-model analysis using Claude, Codex, and Gemini for diverse perspectives

03Context-aware reviews based on spec.md, plan.md, and local coding principles

04Graceful degradation logic ensuring functionality even if specific models are unavailable

05GitHub integration for automated PR commenting and 'Ready for Review' status updates

060 GitHub stars

使用场景

01Automating security and performance audits within the development workflow

02Validating feature implementations against technical specifications before human intervention

03Standardizing PR quality by enforcing specific coding principles across the entire team