Skip to content
↑↓ to move · Enter to openEsc to close
Glossary · Identity

Agent provenance

Updated

When an unfamiliar agent shows up in a phone call or an API request, the first useful question is where it came from. Provenance is the answer in a form that can be checked rather than asserted.

In AgenID, provenance starts with the agent operator: the entity accountable for the agent, which is never the hosting platform.

Definition

Agent provenance is the verifiable record of where an AI agent comes from: which operator declared it, when, and with what stated purpose.

Also known as: agent origin, operator provenance

How it relates to AgenID

The operator signs a ManifestProof that binds the manifest by SHA-256 digest, with a creation time and an operator key whose controller is the agent. Registration adds an append-only ledger event. Anyone can re-verify the chain from the envelope at /verify.

Because the operator key can be fetched from the registry and from the operator's own domain, a verifier can confirm that the key which signed the manifest is the one the operator's domain vouches for, where the operator has published a copy.

Why it matters

A traceable operator gives disputes, complaints and reputation somewhere to land. It also means a verifier can tell when a manifest changed, because any edit changes the digest.

The reference deployment at agenid.com issues nothing above L1_REGISTERED today. L2 and higher require a VerificationAssertion signed by a root authority key, and that key has not been created yet.

What it does not mean

AgenID provenance is about the declaration, not the model or the runtime. v1.1.1 has no object tying an identity to the deployment it executes on, and the platform and configuration_fingerprint manifest keys are reserved. Operator attestations are self-declared and unverifiable by anyone, including AgenID. Provenance also says nothing about what the agent is allowed to do, and it traces the agent, not the content the agent produces.

Frequently asked questions

Does provenance tell me which AI model an agent uses?

No. The manifest does not carry model or platform fields in v1.1.1.

Can provenance be faked?

A different operator can declare its own agent, but it cannot produce another operator's signature. Compare the operator key against the operator's own domain.

Where do I read an agent's provenance?

On its Verification Card or in the JSON envelope.

What agenid.com issues today

The reference deployment at agenid.com issues DECLARED and L1_REGISTERED only. L2_DOMAIN_VERIFIED and higher require a VerificationAssertion signed by a root authority key, and that key has not been created yet. L5 is reserved by the specification and is not issuable at all.

Identity is not permission

Which agent is acting and who declared it is one question. What that agent is allowed to do is another, and it stays with the party on the other side and with the operator. Protocol v1.1.1 defines no signed authorization object.

Give your agent an identity

Register an agent in the browser — the signing key is generated on your device and never sent to AgenID — or resolve an existing AgenID and re-check its signature yourself.