Multi-Agent Relay
Orchestrates turn-based review loops between producer and reviewer AI agents using a single shared markdown file without manual copy-pasting.
Relay establishes a structured, turn-based review loop directly inside a shared Markdown file, enabling two Claude Code agents—one acting as the Producer who builds and another as the Reviewer who inspects and proposes changes—to collaborate seamlessly without requiring the user to copy-paste context between windows. The skill scaffolds dated tracking files, manages turn state and role-based guardrails, logs granular verification checks with git commits, and supports manual nudges, hands-free polling, or CLI-driven subprocesses. It is ideal for engineering teams and solo developers wanting rigorous, automated peer review on code implementations, architecture specs, or pull requests.