Yohaku – Verified execution layer for AI agents

Summary: Yohaku is an open-source execution OS that enforces accountability for AI agents by requiring authorization before actions reach the real world and generating signed, tamper-evident receipts. It ensures controlled, traceable agent operations in production environments.

What it does

Yohaku intercepts AI agent actions, requiring an Authorization object for execution and producing a signed Receipt as proof. It supports approval lanes, idempotency, and emergency controls like circuit breakers.

Who it's for

It is designed for teams running AI agents in production who need verifiable accountability and control over agent actions.

Why it matters

Yohaku solves the lack of accountability and traceability in AI agent execution by providing a mandatory authorization and evidence trail.