Task → Execute
AI-generatedSummary
Executes a natural-language browser task through the selected Browser Use Cloud API version. The supplied UI properties expose task description, starting URL, timeout, structured-output settings, and advanced options for API v2; v3 and v4 use their version-specific task workflows.
Inputs
- API Version (required) — Selects v2, v3, or v4. The default is v2 on node type version 1 and v4 on type version 2.
- Task Description (required) — Natural-language instructions for the browser agent; v2 validates a length of 1–20,000 characters.
- Starting URL — URL where the browser should begin, when provided.
- Timeout — Maximum wait for task completion in seconds (UI default 300; allowed range 10–3600).
- Extract Structured Data — For v2, enables structured extraction using a built-in article, company, contact, or product template, or a custom JSON Schema.
- Advanced Options — For v2, can set the AI model, allowed domains, flash mode, element highlighting, result judging and optional ground truth, maximum steps, metadata, operation vault ID, secrets, an existing session ID, system prompt extension, reasoning visualization, and vision mode. Defaults and choices are shown in the node UI.
Output shape
One n8n item containing task execution data.
The v2 implementation returns task data even when the task reports that it was unsuccessful or incomplete; the result is not limited to successful completion. Structured output is configurable in the supplied properties for v2. The source indicates v4 emulates structured output by parsing returned text while retaining raw text in result, but no structured-output control is present in the supplied operation-scoped properties.