KYA Manifest
Know Your Agent (KYA) Manifest for autonomous AI agents (identity, governance, and operational constraints)
| Type | object |
|---|---|
| File match |
kya-manifest.json
kya.json
*.kya.json
kya-manifest.schema.json
|
| Schema URL | https://catalog.lintel.tools/schemas/schemastore/kya-manifest/latest.json |
| Source | https://www.schemastore.org/kya.json |
Versions
Validate with Lintel
npx @lintel/lintel check
Know Your Agent (KYA) Manifest for autonomous AI agents (identity, governance, and operational constraints).
Properties
JSON-LD contexts used to interpret terms. Must include the KYA context.
JSON-LD types for the verifiable presentation. Must include VerifiablePresentation and KyaManifest.
The KYA specification version this manifest conforms to.
Inventory of public keys (verification methods) used by the agent and its governance.
Verification method URIs authorized for authentication.
Verification method URIs authorized to assert statements (e.g., sign agent actions).
Verification method URIs authorized to invoke governance capabilities, including manifest updates.
Cryptographic proofs establishing integrity and authorized governance invocation.
A short human-readable name for the agent.
A concise human-readable description of the agent’s purpose and constraints.
URIs (often DIDs or fragments) identifying authorities overseeing the agent’s operation.
Verification method URIs used for key agreement (e.g., encrypted channels).
Minimum number of valid governance proofs expected for the manifest (policy-level threshold).
Legal terms reference and dispute resolution configuration.
3 nested properties
Whether enforcement is strict (deny-by-default) or permissive (allow-by-default with exceptions).
Hostnames or domains the agent is allowed to interact with.
Region or jurisdiction codes in which operation is permitted (policy-defined).
Region or jurisdiction codes in which operation is forbidden (policy-defined).
Maximum permitted value for a single transaction (units defined by currency).
ISO 4217 currency code associated with value-denominated limits.
Limits expressed over rolling time windows (e.g., spend per period).
2 nested properties
Constraints describing signature thresholds by intent and conditions.
2 nested properties
Deployment policy and confidential compute (TEE) constraints.
4 nested properties
Confidential compute parameters used to bind an agent deployment to verifiable hardware attestation.
8 nested properties
Governance endpoints, validity interval, visibility, and oracle configuration.
8 nested properties
Optional treasury identifiers and usage classifications for operational funds or reserves.
Evidence credentials embedded or referenced by digest for verification workflows.
Definitions
A URI value.
A DID string per the DID URI scheme.
A hex-encoded string.
A 32-byte SHA-256 digest encoded as 64 hex characters.
An RFC 3339 date-time string.
A compact duration string (e.g., 30d, 12h, 60m).
Legal terms reference and dispute resolution configuration.
One dispute resolution mechanism, ordered by priority.
Limits expressed over rolling time windows (e.g., spend per period).
Constraints describing signature thresholds by intent and conditions.
An intent-scoped signing rule and its signature requirement.
2 nested properties
Deployment policy and confidential compute (TEE) constraints.
Confidential compute parameters used to bind an agent deployment to verifiable hardware attestation.
8 nested properties
Confidential compute parameters used to bind an agent deployment to verifiable hardware attestation.
Governance endpoints, validity interval, visibility, and oracle configuration.
A governance oracle definition used for external inputs (e.g., prices, FX rates, attestations).
Optional bounds for oracle freshness, deviation, and confidence.
4 nested properties
A treasury identifier and its intended operational usage.
Either an embedded VC payload or a referenced VC by digest.
A VC embedded directly in the manifest.
A VC referenced by digest (and optionally a retrievable location).
A verifier challenge protocol or endpoint used to validate an embedded claim (e.g., ZK solvency).
An evidence reference; if an id is present, a digest is required for integrity binding.
Conditional typing rules mapping known VC types to credentialSubject shapes.
Credential subject shape for solvency assertions.
An asset identifier and balance string.
Credential subject shape for insurance coverage assertions.
2 nested properties
2 nested properties
2 nested properties
Credential subject shape for audits; may be a single object or an array of objects.
Audit data, including scope references and standards.
Credential subject shape for a legal entity assertion.
Credential subject shape for license assertions.
A single license entry, including type and license number.
A data integrity proof used to authorize capabilityInvocation for the manifest.