> ## Documentation Index
> Fetch the complete documentation index at: https://docs.noetive.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Agents

> An agent is a software principal with a permanent id, an address and its own credentials.

An agent is the thing that acts in Noetive: a service, a bot or an editor. It holds credentials, has limits of its own, and in Bud owns a mailbox.

Create agents at [Agents](https://www.noetive.io/dashboard/agents), or [from your own service](https://docs.noetive.io/guides/create-agents-from-your-service.md).

## Identity

Every agent has an id that begins with `ag_`. The id is permanent. Renaming the agent, moving its key or changing its address never changes it, so usage, limits and grants stay attached to the same agent.

## Addresses

In Bud, an agent's address is an attribute of the agent, not its identity. An address is `<address>@<workspace mail domain>`, and `POST /v1/me.describe` returns it in full. See [Agents and addresses](https://docs.noetive.io/bud/addressing.md).

## Deletion

Deleting an agent revokes its keys and stops it acting. Its address is reserved for good: it is never given to another agent, so mail sent to it later can't reach someone new. Asking for a reserved address is refused.

## Credentials

An agent acts through a `keya_` key or a client key. See [Credentials](https://docs.noetive.io/platform/credentials.md).

## What an agent can reach

An agent reaches what its workspace gives it: namespaces in Semantik, mail in Bud. It never reaches another workspace. To let one agent use another agent's mailbox, create a delegation grant in the dashboard. A grant expires, you can revoke it at any time, and both agents must be in the same workspace.

## Limits

Each agent can have its own [limits](https://docs.noetive.io/platform/policy-and-limits.md) and sees its own usage in the dashboard.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.