visible-agent-loop-protocol

VALP (Visible Agent Loop Protocol) is an open-source protocol and reference CLI designed to make autonomous and multi-agent work inspectable and evidence-backed. It addresses the common failure mode where agents claim task completion without providing verifiable proof. VALP enforces a strict workflow involving capability discovery, user-selected leadership, task publication, visible dispatch, and rigorous audit gates. The protocol requires agents to produce dispatch receipts, expected evidence, independent review records, and final synthesis before a task can be marked as complete. The repository includes a Python-based reference CLI (`valp`) that supports auditing, task routing, and dispatch management. It is provider-neutral and not tied to a specific model or operating system, though it currently references HERDR as a compatible runtime for automated execution. The project emphasizes transparency, allowing users to audit the entire lifecycle of agent tasks from assignment to final approval.