What it does
codeArbiter is an open-source governance layer for AI coding agents. It lives in the repository and applies hard gates for tests, reviews, security checks and decisions, and keeps durable project context. It works with Claude Code, Codex CLI and Pi.
Use cases
- 01Require tests to pass before an agent's change is accepted
- 02Run security checks on every agent-written change
- 03Keep a written record of technical decisions in the repo