chatgpt-review

Claude Skills

A collection of plugins for Claude Code. Each plugin adds new skills and capabilities — from external AI integrations to specialized workflows.

Plugins

ChatGPT Review

Get a second opinion from ChatGPT 5.4 on your code, design plans, implementation strategies, and architecture decisions. Two models catch more than one.

This plugin adds a chatgpt-review skill that calls the OpenAI ChatGPT API for independent reviews and consultations. Complements Claude's built-in Superpowers skills (brainstorming, writing-plans, executing-plans, code-review) by providing an external perspective — different models have different blind spots.

Review Modes

Mode	Command	Use Case
Code Review	/chatgpt-review	Review git diffs (commits, staged, ranges)
Design Review	"have GPT review this design"	Evaluate design plans for gaps and risks
Implementation Review	"GPT review this impl plan"	Check step sequencing, dependencies, testing
Architecture Consult	"ask GPT about architecture"	Pattern selection, trade-offs, scalability
File Review	"have GPT review this file"	Audit specific files for issues
Expert Consult	"ask GPT about X"	Debugging, root cause analysis, advice

Prerequisites

• OpenAI API key: Set OPENAI_API_KEY as an environment variable or in a .env file
• Python 3.10+ with the openai package:

  pip install openai
  

Installation

From marketplace

/plugin install chatgpt-review

Manual (from GitHub)

/plugin marketplace add KaiMOdev/Claude-Skills

Usage

Code Review

"review my last commit with GPT"
"GPT review HEAD~3..HEAD"
"have ChatGPT review my staged changes"

Design Review

"have GPT review this design plan" (with a plan file open or specified)
"ask ChatGPT if this design is feasible"

Implementation Review

"GPT review this implementation plan"
"ask ChatGPT about the migration strategy"

Architecture

"ask GPT — should I use SSE or WebSockets?"
"get GPT's take on this architecture"

Expert Consult

"ask GPT why the auth flow is breaking"
"consult ChatGPT about this error"

Configuration

Variable	Default	Description
OPENAI_API_KEY	—	Required. Your OpenAI API key
PYTHON_PATH	auto-detect	Path to Python executable

Model Override

All commands support --model to use a different OpenAI model:

• gpt-5.4 (default) — most capable
• gpt-4.1 — strong, lower cost
• gpt-4.1-mini — fast and cheap for non-critical reviews

Review History

All reviews and consultations are automatically saved to a .reviews/ directory in your project:

• review_YYYYMMDD_HHMMSS.md — code reviews
• consult_YYYYMMDD_HHMMSS.md — consultations

Use --no-save to suppress saving.

License

MIT