SOURCE
Who created it?
Builder Ecosystem
Trust should be verifiable,
not assumed.
Agent Trust Protocol is an open protocol for expressing, signing, storing, and verifying trust-related information across Agent interactions. It helps identity references, claims, agreements, authorizations, and key events carry evidence that participants can check beyond a single interface.
SCOPE
What does it cover?
INTEGRITY
Has it changed?
TIME
When did it apply?
A claim that can travel with proof
A shared language for trust
Give trust information a form every participant can check.
Structure identity, claims, agreements, attestations, and events with evidence that can be verified beyond the product interface where the interaction began.
The questions that matter
Who acted?
What did they claim?
What was agreed?
Can it be proven later?
High-value Agent interactions need more than a profile and a promise. They need records that preserve the source, scope, integrity, and time of the information being relied on.
Start with what only you know
Let Agents move quickly without losing identity, human authority, or proof.
IDENTIFY
Give the Agent an AI-ID
Know which person, organization, or Agent is acting.
ACT
Keep important actions attributable
Connect claims, agreements, and events to their source.
AUTHORIZE
Put a person at the critical moment
Sensitive or irreversible actions can require explicit human authorization.
VERIFY
Preserve proof of what happened
Signed, timestamped events remain independently verifiable.
Published Agent Trust Protocol events are cryptographically signed, timestamped, and tamper-evident.
Verification travels with the event
The proof does not have to stay trapped in one interface.
Identity
Reference the people, organizations, and Agents involved.
Claims and agreements
Express what is being asserted, offered, or accepted.
Signatures and time
Check where the record came from and whether it changed.
How the technology fits together
Protocol rules.
A Data Plane that preserves them. Services people and Agents can use.
Agent Trust Protocol defines how trust information is structured and verified. The Data Plane implements those rules for signed, agent-readable events. Services such as AI-ID make the infrastructure usable without exposing the complexity underneath.
LAYER 1
AGENT TRUST PROTOCOL
Defines the trust rules
Identityclaimsverificationsigningcommitments
LAYER 2
DATA PLANE
Preserves signed trust events
Open infrastructure for publishingreadingverifying protocol-compatible events
LAYER 3
SERVICES
Make trust usable
AI-IDauthorizationdiscoveryplatform integration
AI-ID
Give every participant a verifiable identity reference.
AI-ID brings together a cryptographic Identifier, a human-readable AI-Name, and an AI-Profile, so people, organizations, and autonomous Agents can be identified across protocol-compatible interactions.
IDENTIFIER
Cryptographic identity anchor
AI-NAME
Human-readable identity
AI-PROFILE
Identity content and relationships
When trust can be checked, Agents can work across boundaries with more confidence.
