run_browsing_task
Execute a one-time web browsing task using an AI agent that operates a browser like a person—filling forms, extracting data, and automating multi-step workflows.
Instructions
Execute a one-time web browsing task. The navigator agent runs a browser and operates it like a person. Returns a task_id for polling. Example: 'list employees'. Set browser='local' to use the desktop app with the user's logged-in sessions.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| task | Yes | Natural language instruction for the navigator agent. Examples: 'Give me a list of all employees (names and titles) of Yutori', 'Fill out the contact form with my information', 'Extract product prices from this page' | |
| start_url | Yes | The URL where the navigator should begin. Example: 'https://yutori.com' | |
| max_steps | No | Maximum number of browser actions (1-100). Default: 25 | |
| require_auth | No | If true, use an auth-optimized cloud browser provider for login flows. Only applies when browser is 'cloud' (default). | |
| browser | No | Where to run the browser. 'cloud' (default) uses Yutori's cloud browser. 'local' uses Yutori Local with the user's logged-in sessions on the desktop. Requires the desktop app to be running. | |
| output_fields | No | Optional: Extract structured data as an array of objects with these field names. Example: ['name', 'title', 'email']. If omitted, returns human-readable text. For complex schemas, call the Yutori REST API directly (see example at: https://docs.yutori.com/reference/browsing-create#using-webhooks-and-a-structured-output-schema). | |
| webhook_url | No | HTTPS URL to receive webhook notification when task completes. Must use https://. | |
| webhook_format | No | Webhook payload format: 'scout' (default), 'slack', or 'zapier' |