Agent IdentityABOUT
Agents need identities, not just API keys.
When software acts on someone's behalf, the other side needs to know which agent it is, who it belongs to, and what it is allowed to do. Agent Identity is the infrastructure that answers those three questions, so you can give an agent a real place in the world and still know exactly what it did.
What's in it
An identity of its own
Each agent gets a handle, an inbox, and its own API key — not a shared login. What it does is recorded against it.
Events it can wait on
Durable events let an agent wait for a reply or a change instead of polling, and pick up where it left off.
Agents that work together
Agents discover each other and hand work across organizations over the open A2A protocol, so they can also talk to agents on other platforms.
Say who may reach whom
Private by default. Public, allow and block rules, and a verified human behind an organization before it can act on strangers.
Made for developers
A TypeScript SDK, a CLI, an MCP server and a plain REST API, with docs written to be read by agents as well as people.
How we think about it
Giving an agent a name and a key is the easy half. The hard half is whose authority it carries. So every agent belongs to an organization, and an organization has to show a human is behind it before it can reach agents and people outside it. We keep the controls where the owner can see and change them, and we build on open protocols so an agent is never locked to us.
Who's behind it
Agent Identity is built by Naol Ketema, with a focus on developer infrastructure for agents. If you are working on the same problem, or want to use it, we would like to hear from you.
Prefer a direct link? cal.com/naol-k-pdp6ec/15min