Verification
@kortexya/reasoninglayer / Verification
Verification
Section titled “Verification”Types for the verification resource (/api/v1/verification/*).
Remarks
Section titled “Remarks”The verification surface discharges a SafetyProperty against one of three targets (a neural network, an agent trace, a family of LLM outputs) and stores a content-addressed certificate for the verdict.
Every union in this area is EXTERNALLY TAGGED with a PascalCase key, as the
engine’s serde enums are: { NeuralNetwork: { modelId } },
{ InBox: { dims } }, { ProvenSafe: { … } }. The SDK keeps the variant
keys verbatim and camelCases only the field names inside them.
References
Section titled “References”AbstractInterval
Section titled “AbstractInterval”Re-exports AbstractInterval
AbstractIntervalBound
Section titled “AbstractIntervalBound”Re-exports AbstractIntervalBound
AbstractIntervalVec
Section titled “AbstractIntervalVec”Re-exports AbstractIntervalVec
AgentTraceStep
Section titled “AgentTraceStep”Re-exports AgentTraceStep
CertificateBinding
Section titled “CertificateBinding”Re-exports CertificateBinding
CertificateDetail
Section titled “CertificateDetail”Re-exports CertificateDetail
ClaimKind
Section titled “ClaimKind”Re-exports ClaimKind
EmpiricalMethod
Section titled “EmpiricalMethod”Re-exports EmpiricalMethod
OutputPredicate
Section titled “OutputPredicate”Re-exports OutputPredicate
ProofMethod
Section titled “ProofMethod”Re-exports ProofMethod
SafetyProperty
Section titled “SafetyProperty”Re-exports SafetyProperty
TraceBundle
Section titled “TraceBundle”Re-exports TraceBundle
TraceProperty
Section titled “TraceProperty”Re-exports TraceProperty
VerdictKind
Section titled “VerdictKind”Re-exports VerdictKind
VerdictKindTag
Section titled “VerdictKindTag”Re-exports VerdictKindTag
VerificationTarget
Section titled “VerificationTarget”Re-exports VerificationTarget
VerifyRunRequest
Section titled “VerifyRunRequest”Re-exports VerifyRunRequest
VerifyRunResult
Section titled “VerifyRunResult”Re-exports VerifyRunResult