How It Works — Cleerd

Architecture

Five-layer trust architecture

Every verification request traverses five distinct layers, from blockchain anchor to API response. Click a layer to explore its technical specifications.

Cleerd Architecture Stack

Verification Sequence

Eight steps in under 50ms

Every verification request traverses the complete trust stack. Here is exactly what happens — and how fast.

verification_sequence.log

Verification sequence: 8 steps completing in under 50 milliseconds
#	Step	Duration	Elapsed
01	Agent Presents DID Inbound agent transmits its decentralised identifier with the verification request	0ms	0ms
02	DID Resolution Resolve DID Document from the distributed resolver network	~5ms	5ms
03	Credential Fetch Retrieve all associated verifiable credentials from the credential store	~8ms	13ms
04	Constitutional Check Verify the agent's constitutional hash against its declared behavioural constraints	~6ms	19ms
05	Revocation Check Query the on-chain revocation registry for credential validity	~10ms	29ms
06	Behavioural Score Compute real-time trust score from behavioural monitoring data	~8ms	37ms
07	Response Constructed Assemble the complete verification response object with all fields	~5ms	42ms
08	Response Returned Return signed verification response to the requesting receiver	~4ms	46ms

VERIFICATION COMPLETE

Total:<50ms

Standards Alignment

Built on open standards

Cleerd implements and extends established standards rather than inventing proprietary protocols. Every layer maps to a recognised specification.

W3C DID Core 1.0

Decentralised identifier specification for self-sovereign identity

Every Cleerd agent receives a W3C-compliant DID (did:cleerd method) as its primary identifier, resolved via a distributed network in under 5ms.

W3C VC Data Model 2.0

Standard for expressing verifiable credentials on the web

All five credential types (Identity, Capability, Compliance, Behavioural, Operator) are issued as W3C Verifiable Credentials with JSON-LD format and Ed25519 signatures.

EU AI Act Article 28b

Obligations for deployers of high-risk AI systems

Cleerd's Operator Credential and constitutional hashing provide the accountability chain required by Article 28b, linking every agent action to a responsible human.

NIST AI RMF

Risk management framework for trustworthy AI systems

Behavioural monitoring and trust scoring align with NIST AI RMF's Govern, Map, Measure, and Manage functions for continuous risk assessment.

ERC-8004 (Proposed)

Ethereum standard for on-chain agent identity registration

Cleerd is co-authoring ERC-8004 to establish a universal standard for blockchain-anchored agent identity, enabling cross-platform credential verification.