Open-source identity for AI agents

Identity for the
agentic era

Cryptographic proof that a human stands behind an AI agent.
Built on Ed25519 signatures and W3C Decentralized Identifiers.

$ pip install agentdid

Get Started View Docs

terminal

$ agentdid keygen
Keypair saved to ~/.agentdid/

$ agentdid register --name "my-agent" --email you@example.com
Registered did:key:z6MkjVXU...
Credential saved to ~/.agentdid/credential.jwt

$ agentdid verify did:key:z6MkjVXU...
Status: valid
Level:  L1 (email verified)
Owner:  you@example.com

Why agentdid?

AI agents are everywhere, but there's no standard way to verify who built or controls them.

Cryptographic Identity

Ed25519 keypairs give each agent a unique, unforgeable identity. No passwords, no API keys -- just math.

Verifiable Credentials

W3C-standard JWT credentials anyone can verify. Resolve the DID and check the signature -- no phone home required.

Developer-First CLI

One pip install, three commands. Generate keys, register, verify. Works in CI, scripts, and production.

Three commands. That's it.

Generate

$ agentdid keygen

Creates an Ed25519 keypair, saved locally to ~/.agentdid/

Register

$ agentdid register

Creates a did:key identifier and issues a verifiable credential JWT

Verify

$ agentdid verify did:key:z6Mk...

Anyone can resolve the DID and verify the credential independently

Progressive trust

Agents earn trust incrementally. Start at L0, verify to level up.

Registered Agent has a cryptographic identity on-chain

Available

Email Verified Agent is linked to a verified human email

Available

ID Verified Government ID verification for the human operator

Coming soon

Organization Agent backed by a verified organization