Collaborative Human Agent Protocol (CHAP)

The Collaborative Human-Agent Protocol (CHAP) is an open-source protocol that creates a structured, auditable record of interactions between AI agents and humans, capturing both the agent's draft output and the human's edits, rationale, and decisions in a queryable chain. Rather than scattering human-agent decisions across chat threads, code comments, and application logs, CHAP provides a single standardized format for tracking these exchanges, making it possible to reconstruct what happened and why months later.