Writ Protocol

An open delegation-and-evidence protocol for AI agents. Agents act under signed writs that bind scope, intent, and integrity; every action emits a signed receipt.

Writ Protocol is the source layer for agent accountability: a shared, open format that sits above capability primitives and below governance. It does not replace sandboxes or policies — it gives them a common language.

The protocol is in v0. Specification, reference engine, and registry are open source.