# Usage

## Available Actions

LLM Vision exposes five actions:

1. [`image_analyzer`](https://llmvision.gitbook.io/getting-started/usage/image-analyzer): Analyzes image files, image and camera entities
2. [`video_analyzer`](https://llmvision.gitbook.io/getting-started/usage/video-analyzer): Analyzes video files and Frigate events
3. [`stream_analyzer`](https://llmvision.gitbook.io/getting-started/usage/stream-analyzer): Records and analyzes camera entities
4. [`data_analyzer`](https://llmvision.gitbook.io/getting-started/usage/data-analyzer): Updates a sensor value based on images
5. [`create_event`](https://llmvision.gitbook.io/getting-started/usage/create-event): Add a custom event to the LLM Vision Timeline

## Response

`image`, `video`, `stream` and `data` analyzers will return a response value:

```yaml
title: White SUV seen
response_text: >-
  The image shows a white SUV parked on a driveway.
key_frame: /config/media/llmvision/snapshots/0b9ff287-0.jpg
```

{% hint style="info" %}
Depending on your configuration, not all keys may be present. Refer to the parameter reference of each action to learn more.
{% endhint %}

### Accessing the response

To access it, set `response_variable` in your automation or script to e.g. `response`.

To get e.g. the generated title use `{{response.title}}`.

### Using Snapshots in your own Automations

In your own automations use the following template for snapshot paths:

```yaml
{{ response.key_frame.replace('/media', '/media/local') }}
```

Where `response` is the response variable of an analyzer action with `expose_images` enabled.

### Structured Responses

{% hint style="info" %}
Since v1.6.0 LLM Vision supports structured responses (json) for some providers.
{% endhint %}

When `response_format` is enabled and `structure` is provided, some providers can generate a valid JSON object according to the specified JSON schema.

#### Structure Example

```json
{
  "type": "object",
  "properties": {
    "title": {
      "type": "string",
      "description": "Event title"
    },
    "description": {
      "type": "string",
      "description": "Event description"
    },
    "confidence": {
      "type": "number",
      "minimum": 0,
      "maximum": 100
    }
  },
  "required": [
    "title",
    "description",
    "confidence"
  ],
  "additionalProperties": false
}
```

#### Schema Properties

<table><thead><tr><th width="136.4453125">Field</th><th width="146.3828125">Type</th><th width="94.28515625">Required</th><th width="137.39453125">Allowed Values</th><th>Description</th></tr></thead><tbody><tr><td><code>type</code></td><td>string</td><td>Yes</td><td><code>"object"</code></td><td>Specifies the JSON data type the schema describes. In this case, the schema validates a JSON object.</td></tr><tr><td><code>properties</code></td><td>object</td><td>Yes</td><td>Key-value pairs where each key is a field name and each value is a nested schema definition</td><td>Defines the allowed fields of the object and the validation rules for each field.</td></tr><tr><td><code>required</code></td><td>array of strings</td><td>No (but commonly used)</td><td>Array of property names defined in <code>properties</code></td><td>Lists which properties must be present in the JSON object for it to be valid.</td></tr><tr><td><code>additionalProperties</code></td><td>boolean (or schema)</td><td>No</td><td><code>true</code>, <code>false</code>, or a schema object</td><td>Controls whether properties not listed in <code>properties</code> are allowed. <code>false</code> means no extra fields are permitted.</td></tr></tbody></table>


---

# Agent Instructions: Querying This Documentation

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

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://llmvision.gitbook.io/getting-started/usage.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
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.