Hey PH! Stephen here, maker of agentdid. I've been building AI agent infrastructure for the past year and kept hitting the same wall: when two agents interact, or when a user encounters an agent, there's zero way to verify who's behind it. No identity standard. No chain of trust. Just vibes. So I built agentdid — the simplest possible way to give an agent a cryptographic identity tied to a real human. The design principles were: Dead simple DX — pip install, three CLI commands, done. If it takes more than 2 minutes to set up, I failed. No vendor lock-in — Built on W3C DIDs and verifiable credentials. Verification is offline-capable. No phoning home to my server. Progressive trust — Not every use case needs KYC. Start with a keypair (L0), verify email when you need it (L1), add government ID later (L2). Trust scales with stakes. I think agent identity is going to be as foundational as HTTPS was for the web. We're early, but the agents are already here. Would love feedback from anyone building in the agent space. What trust level would you actually need for your use case? What's missing? Star us on GitHub: github.com/Mr-Perfection/agentdid

agentdid — Identity for the agentic era

Open-source identity layer for the agentic era

Open-source identity layer for the agentic era