> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/jo-inc/camofox-browser/llms.txt
> Use this file to discover all available pages before exploring further.

# Get snapshot

> Retrieve the accessibility tree snapshot of the current page with element references

## Endpoint

```
GET /tabs/:tabId/snapshot
```

Returns the current page's accessibility tree as a text snapshot with interactive element references (`e1`, `e2`, etc.) that can be used for clicking and typing.

## Authentication

No authentication required. All endpoints use `userId` for session isolation.

## Path parameters

<ParamField path="tabId" type="string" required>
  The unique identifier of the tab
</ParamField>

## Query parameters

<ParamField query="userId" type="string" required>
  User identifier for session isolation
</ParamField>

<ParamField query="includeScreenshot" type="boolean" default="false">
  Whether to include a base64-encoded PNG screenshot in the response
</ParamField>

<ParamField query="offset" type="integer" default="0">
  Character offset for pagination. Use `nextOffset` from previous response to fetch next chunk.
</ParamField>

<ParamField query="format" type="string" default="text">
  Response format (currently only `text` is supported)
</ParamField>

## Response

<ResponseField name="url" type="string">
  The current URL of the tab
</ResponseField>

<ResponseField name="snapshot" type="string">
  Accessibility tree snapshot with element references in YAML-like format. Interactive elements are annotated with refs like `[e1]`, `[e2]`.
</ResponseField>

<ResponseField name="refsCount" type="integer">
  Total number of interactive element references available
</ResponseField>

<ResponseField name="truncated" type="boolean">
  Whether the snapshot was truncated due to size limits
</ResponseField>

<ResponseField name="totalChars" type="integer">
  Total character count of the full snapshot (before truncation)
</ResponseField>

<ResponseField name="hasMore" type="boolean">
  Whether more content is available beyond the current chunk
</ResponseField>

<ResponseField name="nextOffset" type="integer" optional>
  Character offset to use for fetching the next chunk (only present if `hasMore` is true)
</ResponseField>

<ResponseField name="screenshot" type="object" optional>
  Screenshot data (only present if `includeScreenshot=true`)

  <Expandable title="Screenshot fields">
    <ResponseField name="data" type="string">
      Base64-encoded PNG image data
    </ResponseField>

    <ResponseField name="mimeType" type="string">
      Always `image/png`
    </ResponseField>
  </Expandable>
</ResponseField>

## Pagination

Large pages are automatically truncated to prevent token overflow. The snapshot is limited to 80,000 characters (\~20K tokens), with the last 5,000 characters preserved for pagination/navigation links.

* **First request**: `offset=0` returns the head of the snapshot plus tail (pagination links)
* **Subsequent requests**: Use `nextOffset` value from the previous response
* The tail (last 5K chars) is appended to every chunk so navigation refs remain accessible

## Element references

Interactive elements (buttons, links, textboxes, checkboxes, etc.) are annotated with refs:

```yaml theme={null}
- button "Submit" [e1]
- link "More information..." [e2]
- textbox "Search" [e3]
```

These refs can be used with `/click` and `/type` endpoints. **Refs reset after navigation** - always call `/snapshot` after navigating to get fresh refs.

## Error codes

* `400` - Missing required parameter (`userId`)
* `404` - Tab not found
* `500` - Internal server error

## Examples

### Basic snapshot

```bash theme={null}
curl "http://localhost:9377/tabs/abc123/snapshot?userId=agent1"
```

```json theme={null}
{
  "url": "https://example.com",
  "snapshot": "- heading \"Example Domain\"\n- paragraph \"This domain is for use in illustrative examples...\"\n- link \"More information...\" [e1]\n",
  "refsCount": 1,
  "truncated": false,
  "totalChars": 156,
  "hasMore": false,
  "nextOffset": null
}
```

### Snapshot with screenshot

```bash theme={null}
curl "http://localhost:9377/tabs/abc123/snapshot?userId=agent1&includeScreenshot=true"
```

```json theme={null}
{
  "url": "https://example.com",
  "snapshot": "- heading \"Example Domain\"...",
  "refsCount": 1,
  "truncated": false,
  "totalChars": 156,
  "hasMore": false,
  "nextOffset": null,
  "screenshot": {
    "data": "iVBORw0KGgoAAAANSUhEUgAAA...",
    "mimeType": "image/png"
  }
}
```

### Paginated snapshot

```bash theme={null}
# First chunk
curl "http://localhost:9377/tabs/abc123/snapshot?userId=agent1"
```

```json theme={null}
{
  "url": "https://longpage.com",
  "snapshot": "[content]\n[... truncated at char 75000 of 120000. Call snapshot with offset=75000 to see more. Pagination links below. ...]\n[tail content with pagination links]",
  "refsCount": 245,
  "truncated": true,
  "totalChars": 120000,
  "hasMore": true,
  "nextOffset": 75000
}
```

```bash theme={null}
# Next chunk
curl "http://localhost:9377/tabs/abc123/snapshot?userId=agent1&offset=75000"
```

## Caching behavior

For `offset > 0` requests, the server uses a cached snapshot from the last `offset=0` call. This ensures consistent pagination without rebuilding the accessibility tree. The cache is invalidated after navigation or interaction.

## Excluded elements

The following elements are excluded from snapshots to avoid interference:

* Comboboxes (date pickers, complex dropdowns)
* Elements matching patterns: `date`, `calendar`, `picker`, `datepicker`

This prevents accidental triggering of complex widgets during automation.