Agentic CPL — Cryptographic Trust Layer (TRUST-ANCHOR™) · Live Simulation

AI agent vs. physical actuators, with and without an enforcement layer Network: Online

Without Trust Layer

AI agent ──▶ Actuators

With Agentic CPL

AI agent ──▶ Enforcement layer ──▶ Actuators
Proposed
0
Passed
0
Clamped
0
Rejected (unsigned)
0
Vector received → block decision
no clamp yet
In-browser simulation, measured with performance.now() (timer resolution is browser-limited). Network requests made by this page: 0

Enforcement log one line per command or event, newest first

time· proposed vector · result · signature (12 hex) · actuator

Immutable policy read-only (frozen object)


        
SHA-256: computing…

      

      

        
Tamper-evident audit log SHA-256 hash chain

        

          
          
          
        

        

        

      

    

  


  

    
Production architecture target design · not running in this demo

    
      
      
        
        AI agent
        CPU / NPU
        not trusted: may drift, hallucinate
        or be injected
        
        vectors
        
        Enforcement layer
        isolated MCU / ARM TrustZone
        signing keys in TPM 2.0
        policy check · clamp · sign
        
        signed
        
        Actuators
        relays · ESCs · flight controller
        accept signed commands only
      
    
    
Stage 3 of the TRUST-ANCHOR specification (hardware root of trust, firmware attestation) is shown as the target architecture only. This simulation does not implement a TPM or attestation.

  






  
  
Simulation running in your browser. In production, the policy engine runs in an isolated execution environment (MCU / ARM TrustZone) with signing keys held in TPM 2.0. Cryptography in this demo: ECDSA P-256 and SHA-256 via WebCrypto.