> For the complete documentation index, see [llms.txt](https://docs.umbraco.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.umbraco.com/ai-in-umbraco/19.latest/add-ons/agent/api.md).

# API

Management API endpoints for the Agent add-on.

The Agent Management API provides endpoints for creating, managing, and running AI agents.

## Endpoints

| Method | Endpoint                                                                                            | Description                                        |
| ------ | --------------------------------------------------------------------------------------------------- | -------------------------------------------------- |
| GET    | [`/agents`](/ai-in-umbraco/19.latest/add-ons/agent/api/list.md)                                     | List all agents                                    |
| GET    | [`/agents/{agentIdOrAlias}`](/ai-in-umbraco/19.latest/add-ons/agent/api/get.md)                     | Get an agent by ID or alias                        |
| POST   | [`/agents`](/ai-in-umbraco/19.latest/add-ons/agent/api/create.md)                                   | Create a new agent                                 |
| PUT    | [`/agents/{agentIdOrAlias}`](/ai-in-umbraco/19.latest/add-ons/agent/api/update.md)                  | Update an existing agent                           |
| DELETE | [`/agents/{agentIdOrAlias}`](/ai-in-umbraco/19.latest/add-ons/agent/api/delete.md)                  | Delete an agent                                    |
| GET    | [`/agents/{alias}/exists`](/ai-in-umbraco/19.latest/add-ons/agent/api/list.md#check-alias-exists)   | Check if an agent alias exists                     |
| GET    | [`/agents/surfaces`](/ai-in-umbraco/19.latest/add-ons/agent/api/list.md#list-surfaces)              | List all registered agent surfaces                 |
| GET    | [`/agents/workflows`](/ai-in-umbraco/19.latest/add-ons/agent/api/list.md#list-workflows)            | List all registered agent workflows                |
| POST   | [`/agents/{agentIdOrAlias}/run`](/ai-in-umbraco/19.latest/add-ons/agent/api/run.md)                 | Run an agent (non-streaming JSON response)         |
| POST   | [`/agents/{agentIdOrAlias}/stream`](/ai-in-umbraco/19.latest/add-ons/agent/api/stream.md)           | Stream agent updates (SSE)                         |
| POST   | [`/agents/{agentIdOrAlias}/stream-agui`](/ai-in-umbraco/19.latest/add-ons/agent/api/stream-agui.md) | Stream agent events using the AG-UI protocol (SSE) |

## Base URL

```
/umbraco/ai/management/api/v1
```

## Agent Object

{% code title="Agent" %}

```json
{
    "id": "3fa85f64-5717-4562-b3fc-2c963f66afa6",
    "alias": "content-assistant",
    "name": "Content Assistant",
    "description": "Helps users write and improve content",
    "agentType": "standard",
    "profileId": "d290f1ee-6c54-4b01-90e6-d701748f0851",
    "guardrailIds": [],
    "surfaceIds": ["copilot"],
    "scope": null,
    "config": {
        "$type": "standard",
        "contextIds": ["e401f2ff-7d65-5c12-a1f7-e812859a1962"],
        "instructions": "You are a helpful content assistant...",
        "allowedToolIds": [],
        "allowedToolScopeIds": ["content-read", "search"],
        "outputSchema": null,
        "userGroupPermissions": {}
    },
    "isActive": true,
    "dateCreated": "2024-01-15T10:30:00Z",
    "dateModified": "2024-01-20T14:45:00Z",
    "version": 2
}
```

{% endcode %}

## Properties

| Property       | Type      | Description                                                      |
| -------------- | --------- | ---------------------------------------------------------------- |
| `id`           | guid      | Unique identifier                                                |
| `alias`        | string    | Unique alias for code references                                 |
| `name`         | string    | Display name                                                     |
| `description`  | string    | Optional description                                             |
| `agentType`    | string    | `standard` or `orchestrated`                                     |
| `profileId`    | guid      | Associated AI profile (null uses default)                        |
| `guardrailIds` | guid\[]   | Guardrail IDs for safety and compliance checks                   |
| `surfaceIds`   | string\[] | Surface IDs for categorization                                   |
| `scope`        | object    | Optional scope defining where the agent is available (see below) |
| `config`       | object    | Type-specific configuration (see below)                          |
| `isActive`     | bool      | Whether the agent is available                                   |
| `dateCreated`  | datetime  | When the agent was created (UTC)                                 |
| `dateModified` | datetime  | When the agent was last modified (UTC)                           |
| `version`      | int       | Current version number                                           |

### Scope (`scope`)

Defines where the agent is available using allow and deny rules. When `null`, the agent is available in all contexts.

| Property     | Type    | Description                                           |
| ------------ | ------- | ----------------------------------------------------- |
| `allowRules` | rule\[] | Rules where the agent is available (OR between rules) |
| `denyRules`  | rule\[] | Rules where the agent is denied (takes precedence)    |

Each rule has the following properties (AND between properties, OR within an array):

| Property      | Type      | Description                                  |
| ------------- | --------- | -------------------------------------------- |
| `sections`    | string\[] | Section pathnames (e.g., `content`, `media`) |
| `entityTypes` | string\[] | Entity types (e.g., `document`, `media`)     |

### Standard Config (`config` when `agentType` is `standard`)

| Property               | Type      | Description                                             |
| ---------------------- | --------- | ------------------------------------------------------- |
| `$type`                | string    | Always `"standard"`                                     |
| `contextIds`           | guid\[]   | AI Contexts to inject                                   |
| `instructions`         | string    | Agent system prompt                                     |
| `allowedToolIds`       | string\[] | Explicit tool permissions                               |
| `allowedToolScopeIds`  | string\[] | Scope-based tool permissions                            |
| `outputSchema`         | object    | Optional JSON Schema constraining agent output          |
| `userGroupPermissions` | object    | Per-user-group permission overrides (keyed by group ID) |

### Orchestrated Config (`config` when `agentType` is `orchestrated`)

| Property     | Type   | Description                       |
| ------------ | ------ | --------------------------------- |
| `$type`      | string | Always `"orchestrated"`           |
| `workflowId` | string | ID of the registered workflow     |
| `settings`   | object | Workflow-specific settings (JSON) |

## Related

* [Agent Concepts](/ai-in-umbraco/19.latest/add-ons/agent/concepts.md)
* [Streaming](/ai-in-umbraco/19.latest/add-ons/agent/streaming.md)


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.umbraco.com/ai-in-umbraco/19.latest/add-ons/agent/api.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.