Verifiable Execution
If my agent reuses a cached model response, does the verifiable record reflect that the response came from a cache?
Yes, Catalyst includes a cache-hit indicator in the step metadata when a cached response is used. The record remains signed and verifiable, but it only proves that the step used a cached value, not that the cache was fresh. The auditor sees the provenance identifier, which may be a pointer to the original step’s hash. Moreover, if the original step is later invalidated due to model drift, the cache hit does not automatically become invalid; verification still succeeds.
Was this article helpful?
Your feedback helps improve Diagrid's FAQ experience.
Keep reading
More Diagrid FAQ articles
- Verifiable Execution
What is a verifiable execution record in Catalyst?
A signed, immutable sequence of agent steps enabling auditors to prove run integrity through independent replay. It does not validate decision correctness.
- Verifiable Execution
How does Catalyst ensure that step records are tamper-evident?
Catalyst uses a hash chain and digital signatures per step, so any alteration breaks the chain and becomes detectable through signature verification.
- Verifiable Execution
Can an auditor independently replay a Catalyst agent run to verify its outcome?
Auditors can replay runs using signed step records and initial state to verify output consistency, though non-determinism may limit exact reproduction.