# Agent Identity

> A way to identify an AI agent as an actor with its own metadata, credentials, permissions, and delegation record.

- Category: Tools & Protocols
- Stage: Rising
- Also called: AI agent identity, Agent principal
- Canonical: https://www.joelcsaji.com/ai-dictionary/agent-identity

## Plain English

The system should know which agent acted, not just which human account or API key was nearby.

## Example

A deployment agent has a short-lived identity bound to one repository, one task, and one approval chain.

## Why it matters

Agent identity is becoming necessary for authorization, audit logs, revocation, and agent-to-agent delegation.

## Related

- [A2A](https://www.joelcsaji.com/ai-dictionary/agent2agent)
- [Capability Token](https://www.joelcsaji.com/ai-dictionary/capability-token)
- [Approval Gate](https://www.joelcsaji.com/ai-dictionary/approval-gate)

## Sources

- [NIST: AI Agent Identity and Authorization](https://www.nccoe.nist.gov/sites/default/files/2026-02/accelerating-the-adoption-of-software-and-ai-agent-identity-and-authorization-concept-paper.pdf)
- [arXiv: Agent Identity Protocol](https://arxiv.org/abs/2603.24775)