CAVA, or Canonical Action Verification and Attestation, proposes a runtime-semantics layer that converts heterogeneous agent activity into canonical action objects. It targets governance questions such as what was approved, whether approval is bound to execution, and whether an independent verifier can reproduce the same action identity. The paper distinguishes CAVA from Proof-Carrying Agent Actions (PCAA): PCAA defines the deployer-owned route-review-prove process, while CAVA defines the stable action object governed by that process. A reference implementation is evaluated with 96 seeds and 384 variants across semantic equivalence, separation, wrapper bypass, approval binding, receipt reproducibility, tamper detection, portability, policy degradation, and Azure deployment drills.
No heat snapshots are available in the last 24 hours.