Your AI agent called a tool. Can you prove it followed the rules?

Short summary

AI agents need provable audit trails showing policies were enforced before tools executed. WasmAgent's ActionEvidence Protocol records pre/post state digests, policy decisions, and budget tracking alongside tool calls, with OpenTelemetry integration for real-time observability. Essential for multi-agent systems where delegation chains require proof of authorization scope.

• •Logs alone cannot prove policy gates ran before tool execution
• •ActionEvidence Protocol provides Zod-validated audit records with state digests and capability decisions baked in
• •Supports multi-agent delegation with scope leases and authorization chains

Generated with AI, which can make mistakes.