All activity
koyama_shoheileft a comment
I built Yohaku because I kept seeing the same problem: AI agents executing in production with no accountability trail. Who authorized it? What data did it access? Can you stop it if something goes wrong? Yohaku sits between the AI agent and the real world. Every action requires an Authorization object before execution — then leaves a signed, tamper-evident Receipt. The core flow: Plan → Approve...
YohakuVerified execution layer for AI agents
AI agents are executing in production with no accountability.
Yohaku is the verified execution OS for agent fleets.
Every action passes through Yohaku before it touches the real world.
Plan → Approve → Confirm → Receipt (signed, tamper-evident proof)
Key features:
• Approval Lanes (Auto / Review / Gate)
• KYA: Know Your Agent
• Evidence Pack
• Freeze / Circuit Breaker
• PROD Idempotency (no double execution)
• Context Receipt
Human optional. Accountability mandatory.
YohakuVerified execution layer for AI agents