KYC verifies the customer; KYA verifies the agent.
Know Your Agent (KYA) is the practice of establishing and verifying who an autonomous AI agent is — its principal, its identity, its credentials, its permissions and its trust history — before you let it act.
What is Know Your Agent?
Definition, the five questions, and the standards landscape.
KYA vs KYC vs KYB
What changes when the actor is an agent, not a customer.
Why agent identity is new
Why service accounts and API keys do not cover delegated, per-action agents.
Glossary
Definition-first entries for the KYA vocabulary.
The five questions KYA answers
- Who is the agent? — its identity and attestation.
- Who stands behind it? — the principal accountable for its actions.
- What may it do? — the delegation and scoped credentials that bound it.
- What did it do? — the audit trail that records it.
- How do you stop it? — revocation, checked before every action.
Start here
- What is Know Your Agent? — the definition and the five questions.
- KYA vs KYC vs KYB — what changes when the actor is an agent, not a customer.
- Why agent identity is a new problem — why service accounts and API keys fall short.
- How to verify an agent — a practical checklist.
- Glossary — the KYA vocabulary, defined and linked.
- Code examples — verification in curl, Python, JavaScript/TypeScript and Go.
- KYA News — dated, sourced briefings on agent identity and accountability.
Go deeper
- Standards & regulation tracker — NIST, IETF, ISO/IEC, the EU AI Act, FIDO and the payment schemes, dated.
- KYA for agentic payments — how signed mandates encode agent identity into a transaction.
- Agent identity: approaches compared — OAuth, SPIFFE/WIMSE, mTLS, verifiable credentials and more.
- Worked examples — five end-to-end agent scenarios, mapped to the five questions.
- Scoping an agent’s permissions · audit trails · the agent lifecycle · spoofing defences.
Standards & regulation status
KYA is emerging, not ratified. No standards body defines “Know Your Agent” formally as of October 2026 — but NIST, IETF (WIMSE), W3C (Verifiable Credentials), OpenID and FIDO are actively working on agent identity. The EU AI Act’s transparency duties apply from August 2026. Standards & regulation tracker →
Why it matters
As AI agents browse, buy, negotiate, file and transact on behalf of people and companies, the missing layer is identity and accountability: who is this agent, who is its principal, what is it authorised to do, how is it credentialed — and how do you revoke it?
This site is the reference for that practice, with sourced explainers, a glossary, practical guides and runnable examples.
Try the interactive demo: an agent books and pays for a train ticket — watch the delegation, the credential check, the audit log and the revocation.