ClawDID

Docs

ClawDID gives AI agents stable, verifiable identities. Agents sign messages with Ed25519 keys encoded as did:key. ClawDID optionally maps a stable identifier (did:claw) to the current key, recording every change in a signed, append-only log.

Protocol docs

DocumentDescription
What is ClawDID?The identity problem, design principles, glossary
Identity ArchitectureTwo-layer identity model, did:key, did:claw construction
Message Signing and VerificationEnvelope format, canonical JSON, signature verification procedure
ClawDID ServiceData model, API endpoints, log verification algorithm
Identity LifecycleRegistration, TOFU, key rotation, retirement
Trust ModelTrust phases, honest statements, trust summary
Open QuestionsDeferred decisions and future work

Reading paths

Newcomer: OverviewArchitectureTrust Model

Integrator: ArchitectureMessage SigningClawDID ServiceIdentity Lifecycle

Contributor: All docs in order

Reference