Does this skill require external software?

Yes, you must have the @openai/codex or @google/gemini-cli packages installed globally via npm to use their respective review capabilities.

Can I review specific branches or commits?

Absolutely. The skill supports reviewing uncommitted changes, comparing your current branch against the default branch (main/master), or analyzing a specific commit SHA.

How does it handle project-specific coding standards?

It automatically detects CLAUDE.md or AGENTS.md files in your repository root and provides them as context to the external models to ensure reviews align with your project standards.

What happens if the diff is very large?

For diffs exceeding 2,000 lines, the skill issues a warning about potential reasoning quality and suggests narrowing the scope to specific files or directories.

Is the Gemini CLI integration safe to use?

Gemini is invoked with the --yolo flag for headless operation, meaning it will execute tool calls requested by its extensions without a separate confirmation prompt.

Second Opinion Code Review

Name: Second Opinion Code Review
Author: plurigrid

byplurigrid

•

협업 도구

Integrates external LLM reviews from OpenAI Codex and Google Gemini into your local development workflow for independent code audits.

The Second Opinion skill allows developers to obtain cross-model perspectives on their code by piping diffs into specialized external CLIs like OpenAI Codex and Google Gemini. It automates the process of gathering context, detecting diff scopes (uncommitted, branch-based, or specific commits), and running focused audits for security or performance. By providing an independent layer of analysis alongside Claude's native capabilities, it helps identify edge cases and bugs before code reaches the pull request stage.

주요 기능

01Multi-LLM comparative reviews using OpenAI Codex and Google Gemini CLIs

02Parallel tool execution for faster results and side-by-side comparisons

03Flexible diff scoping for uncommitted changes, branch comparisons, or specific commits

048 GitHub stars

05Automated security auditing and dependency manifest scanning

06Context-aware analysis using project standards from CLAUDE.md or AGENTS.md

사용 사례

01Performing a final security and logic sanity check before opening a Pull Request

02Getting an independent second opinion on complex refactors from different AI models

03Automating dependency vulnerability scans when manifest files are modified

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi second-opinion

For use in Claude.ai and ChatGPT

Download Skill