ai-4-phi/second-opinion

This is a plugin for Claude Code, an AI coding assistant. The plugin lets Claude ask another AI model to review its work, like getting a second opinion from a colleague. You can prompt it to check a piece of code, a plan, or a draft by typing a simple command. Claude then reads the feedback and decides what to do with it. It can also trigger these reviews on its own when it thinks a task deserves an outside check. The plugin supports several external AI providers: Kimi, Gemini, OpenAI, DeepSeek, xAI, GLM, and MiniMax. You can pick which one you want for each review, or let the plugin choose based on which API keys you have configured. The default provider is Kimi. Each review sends your content to exactly one of these services, and the README is clear about the privacy tradeoff. Your material goes to the third party you select, under their data policies, so you should not send confidential work to a provider you would not trust with it directly. Installing the plugin is done through the ai4phi marketplace. You need Python 3 on your machine, but no extra packages. You also need an API key for at least one provider, saved in your environment so Claude Code can find it. The plugin works on macOS and Linux. Windows is untested, though it should work through the Windows Subsystem for Linux. Each review costs a small amount of money, billed to your API key. Some providers, like DeepSeek and Gemini Flash, are nearly free, while others cost a few cents per review. The README includes a troubleshooting section covering common issues, such as missing API keys, failed requests, and a note about large requests being handled differently. It also points to additional documentation for deeper details on usage and provider behavior. The project is MIT licensed. The full README is longer than what was shown.