AgentAdmit is the authorization layer for the agent economy. It gives people a safe way to let an AI agent access their account: the user grants the agent specific, time-limited access, the credential is delivered to the user instead of flowing to the agent through an automated channel, and every grant is scoped, revocable, and auditable.
Why it exists
Most AI agent access today runs on API keys, OAuth redirects, or environment variables. Those channels move a credential with no human at the point of delivery, so a prompt injection or a misconfiguration can intercept it. AgentAdmit removes that automated path. The credential goes to the human, the human hands it to their agent, and the authorization decision happens outside the agent’s execution context, so an injected instruction has no way to escalate access.
What this blog is for
This is where we work through the real questions builders are asking: how to let an agent into a user account without handing over the keys, what user-mediated authorization means, whether an agent can approve its own permission request, whose consent applies when a person, the app’s own AI, and an outside agent reach the same data, and how MCP security and prompt injection change the threat model. The through-line is keeping the human in control. Practical and specific. The terms we rely on are defined in the AI agent authorization glossary.
The company
AgentAdmit is patent-pending infrastructure for user-mediated authorization of AI agents. The model is also documented in an Internet-Draft submitted to the IETF OAuth community (a work in progress, not an adopted standard). To see how it works or build with it, visit agentadmit.com.
Who writes this
Every post here is written by Christopher Emerson, the founder of AgentAdmit and the named inventor on its patent-pending user-mediated authorization model.
Contact
Questions, corrections, or topic ideas: christopher@agentadmit.com.