codex-second-opinion

Audit Claude's output for bugs, dead code, and architectural risks via Codex.

12|Updated Apr 28, 2020

Search Tags:#architecture #audit #codex #claude #code-review #read-only #bug-hunting

Ready to run

npx skills add https://github.com/gideonshaked/dotfiles --skill codex-second-opinion-gideonshaked

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: codex-second-opinion
Source: https://github.com/gideonshaked/dotfiles/tree/main/agents/claude/skills/codex-second-opinion
Command: npx skills add https://github.com/gideonshaked/dotfiles --skill codex-second-opinion-gideonshaked

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Audits Claude's work using Codex to identify mistakes, gaps, and architectural issues, providing a trusted second opinion.

Core Features & Use Cases

Read-only Codex audit of Claude's outputs to detect bugs, dead code, and design flaws.
Supports focus areas or file-path scoping to narrow the audit.
Produces an unembellished audit report for user review.

Quick Start

Ask Codex to audit Claude's latest work with an optional focus area or file path

Frequently Asked Questions about codex-second-opinion

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I get a second opinion on code generated by Claude?▼

You can audit Claude's output by running Codex in a read-only sandbox. This second opinion identifies critical bugs, dead code, and architectural risks, returning an unembellished plain text report for your review.

Can I scope a code audit to a specific file path or focus area?▼

Yes, you can scope the audit by providing a specific focus area or file path. If none is supplied, the mechanism performs a broad, project-wide audit of the output instead.

What kind of architectural risks and bugs does a cross-LLM code review find?▼

A cross-LLM code review identifies critical bugs, dead code, architectural risks, and quality gaps in generated code. It produces an unembellished report so you can catch mistakes and design flaws.

Does the Codex audit run in a read-only sandbox?▼

Yes, the Codex audit operates in a read-only sandbox to safely analyze the code. This ensures the review process identifies mistakes and gaps without modifying your original files.

How do I perform a project-wide code review for dead code and quality gaps?▼

To perform a project-wide code review, initiate the audit without specifying a focus area or file path. This triggers a broad scan that identifies dead code, quality gaps, and architectural risks.