chorus-codes/chorus

Chorus sits between your existing CLI workflow and the moment you commit, convening 2-4 different LLMs to review a code decision before it goes out. The positioning is sharp: it does not replace your primary AI coding assistant, it stress-tests the assistant's output by routing the same decision through models with different biases and training cutoffs. The thing that makes this worth your Saturday is the peer-review framing rather than the consensus framing — the goal is disagreement surfacing, not averaging. That is a real differentiator over just running the same prompt twice. Honest reservation: the value compounds only when the models actually disagree, and for boilerplate decisions they often do not, so the overhead may feel heavy for routine commits. Where it shines is architectural decisions, API-design choices, and security-adjacent code. The bring-your-own-CLI design means zero lock-in, which is the right call. -> Best for: solo founder or SaaS team of 2-5 shipping production code without a human reviewer on standby