ai-jury

ai-jury is a command-line tool by Berkay Turancı (GitHub berkayturanci). It convenes coding agents and model APIs from different vendors, such as Claude and Codex, into a review "jury" that debates a diff, pull request or issue over several rounds and then writes one review report. It can also run in CI as a gate before merging. It isn't a model itself; the reviewing is done by the AI tools you've installed.

Who it's for:Dev teams that want several AIs to check each other and add one more layer of code review. Readers who don't write code won't get much from it.

Before you start: You need Python 3.11 or later; install with brew, pipx or pip. The README also offers curl https://ai-jury.dev/install.sh | sh, which runs a remote script directly. You install and sign in to the AI CLIs you want to use yourself. Docs and interface are mostly in English. The author's security notes say that in default review mode, Claude gets no tools and Codex runs read-only, though Codex can still read any file your user account can. If you opt in, one adapter runs with --dangerously-skip-permissions and can read, write and go online; commands that change code, like implement and fix, also need --allow-write. With --post or in CI, it uses your gh identity to post comments and labels on PRs or issues. If you use these features, weigh the safety risks yourself. AI Inspiration World is not responsible for any consequences. The tool is open source; AI CLIs and model APIs may cost money. MIT license.

Based on the project's public documentation. Not tested by AI Inspiration World.