Summary

Chorus addresses the challenge of subtle errors in AI-generated code by enabling multi-LLM peer review. It targets developers who use AI coding assistants, offering a way to catch AI blind spots before shipping code. The primary benefit is enhanced code quality and reduced risk of production bugs, achieved at zero additional cost by leveraging existing AI subscriptions.

How It Works

Chorus orchestrates 2-4 different AI LLMs to review code changes in parallel, requiring unanimous agreement before approval. It leverages users' existing AI subscriptions (e.g., Claude Pro, ChatGPT Plus, Gemini Advanced) by driving their respective CLIs headlessly. This local-first approach ensures code never leaves the user's machine, and by using diverse AI vendors, it mitigates the risk of shared biases and blind spots inherent in single-model reviews.

Quick Start & Requirements Installation involves npm i -g chorus-codes (requires Node.js >= 20). Prerequisites include at least one AI CLI (e.g., @anthropic-ai/claude-code, @openai/codex, @google/gemini-cli) or an OpenRouter API key. Setup proceeds with chorus init to detect and wire up AI tools, followed by chorus start --ui to launch the web interface. Further details are available on the website (https://chorus.codes), roadmap ([./ROADMAP.md](https://github.com/chorus-codes/chorus/blob/main/ROADMAP