What is the benefit of the consensus voting model?

By comparing results from three distinct AI architectures, the MAGI system reduces individual model hallucinations and provides a more objective, balanced evaluation of code health.

How does the MAGI Review skill work?

It launches three AI models—Claude, Gemini, and Codex—simultaneously to analyze your code from different perspectives: logic, quality, and security. It then merges their findings into a single, comprehensive consensus report.

Can I use it on specific files or full Pull Requests?

Yes, the /magi-review command is versatile and supports individual file paths, directories, or specific Pull Request identifiers.

Does it require separate API keys for Gemini and Codex?

Yes, this skill requires the corresponding MCP tools (mcp__gemini and mcp__codex-cli) to be configured and authorized in your Claude Code environment.

MAGI Parallel Code Review

Name: MAGI Parallel Code Review
Author: weiting-tw

byweiting-tw

•

Seguridad y Pruebas

Conducts multi-engine parallel code reviews using Claude, Gemini, and Codex to provide consensus-driven feedback on logic, quality, and security.

MAGI Parallel Code Review is a sophisticated Claude Code skill inspired by the MAGI supercomputer architecture. It leverages a triple-engine approach—MELCHIOR-1 (Claude Sonnet), BALTHASAR-2 (Gemini Flash), and CASPER-3 (Codex)—to simultaneously analyze codebases or pull requests. By assigning specialized roles to each engine, such as architectural logic, naming consistency, and OWASP security auditing, the skill aggregates independent findings into a unified consensus report. This multi-perspective validation ensures high-reliability audits, minimizes false positives, and identifies complex vulnerabilities that a single model might overlook.

Características Principales

01Automated consensus-based reporting for high-confidence feedback

02Seamless integration with file paths and Pull Request references

03Parallel triple-engine analysis using Claude, Gemini, and Codex

042 GitHub stars

05Comprehensive logic and architectural consistency validation

06In-depth security auditing focusing on OWASP Top 10 vulnerabilities

Casos de Uso

01Automating multi-model code quality checks within a pull request workflow

02Resolving complex architectural debates through objective AI consensus

03Performing rigorous security audits before merging critical production code

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add weiting-tw/magi-system magi-review

For use in Claude.ai and ChatGPT