Overview
This node, named "Browser Use," enables users to automate any web task using natural language instructions processed by AI agents. It is designed to perform complex browsing and interaction tasks on websites without requiring manual scripting. Users describe what they want the AI agent to do in plain English, optionally provide a starting URL, and configure advanced options such as timeout and maximum steps.
Common scenarios include:
- Automating data extraction from websites.
- Filling out forms or interacting with web pages automatically.
- Searching for information online and summarizing results.
- Extracting structured data like product details, company info, or contact information from web content.
Practical example:
- Task: "Go to Google and search for browser automation."
- Starting URL: "https://www.google.com"
- The AI agent will navigate to Google, perform the search, and return relevant results or extracted data.
Properties
| Name | Meaning |
|---|---|
| Task Description | Describe the web task you want the AI agent to perform in natural language (1-20,000 characters). Example: "Go to Google and search for browser automation" or "Fill out the contact form with my details". |
| Starting URL (Optional) | Optional URL where the task should start. Must be a valid HTTP/HTTPS URL. Example: "https://example.com". |
| Timeout (Seconds) | Maximum time in seconds to wait for the task to complete. Range: 10 to 3600 seconds (1 hour). Default is 300 seconds. |
| Extract Structured Data | Boolean flag to enable extraction of data in a specific JSON format for easier processing. |
| Configure the data structure you want to extract below ↓ | Informational notice shown when structured output is enabled. |
| Data Template | Choose a pre-built JSON schema template for structured data extraction or select "Custom Format" to define your own JSON schema. Options include: Article/Blog Content, Company Information, Contact Information, Product Information, Custom Format. |
| 💡 Using a pre-built template will automatically extract the fields shown above. To see the full JSON schema or create your own structure, select "Custom Format". | Informational notice about templates. |
| Custom Data Format (JSON Schema) | Define a custom JSON schema structure for the AI to extract when "Custom Format" is selected. Must be a valid JSON schema object describing the expected output structure. |
| Advanced Options | Collection of additional settings: - Max Steps: Maximum number of steps the AI agent can take (1-200). Default 30. - AI Model: Select which AI model to use for the task. Default is "Gemini 2.5 Flash". |
| Need an API key? Sign up at cloud.browser-use.com | Notice with link to sign up for an API key required to use this node. |
| Documentation: docs.cloud.browser-use.com | Notice with link to official documentation for further reference. |
Output
The node outputs JSON data representing the result of the automated web task. The structure depends on whether structured output is enabled:
If structured output is disabled, the output contains general task status and messages, including:
status: Current status of the task (e.g., "finished", "stopped").isSuccess: Boolean indicating if the AI agent successfully completed the task.agentMessage: Human-readable message about the task outcome.cloudUrl: URL to view the session in the Browser Use cloud interface (if available).- Other metadata related to the task execution.
If structured output is enabled, the output includes a JSON object strictly following the selected or custom JSON schema. This allows easy parsing and integration of extracted data such as articles, company info, contacts, products, or any user-defined format.
The node does not output binary data.
Dependencies
- Requires an API key credential for the Browser Use service (an external web automation AI platform).
- The node makes HTTP requests to the Browser Use API endpoints to create and monitor tasks.
- Network connectivity to the Browser Use API service is necessary.
- No other external dependencies are required.
Troubleshooting
Common Issues
- Empty or too long task description: The task description must be non-empty and no longer than 20,000 characters.
- Invalid starting URL: If provided, the URL must be valid and start with http:// or https://.
- Timeout out of range: Timeout must be between 10 and 3600 seconds.
- Invalid JSON schema: When structured output is enabled with a custom schema, the JSON schema must be valid and correctly formatted.
- API authentication failure: Occurs if the API key is missing or invalid.
- Rate limiting: Too many concurrent sessions or requests may cause rate limit errors.
- Task stopped unexpectedly: The task might be halted due to errors during execution.
Error Messages and Resolutions
"Task description is required and cannot be empty": Provide a valid task description."Task description must be 20,000 characters or less": Shorten the task description."Invalid URL format. Please provide a valid URL starting with http:// or https://": Correct the starting URL."Timeout must be between 10 and 3600 seconds (1 hour)": Adjust the timeout value within allowed range."Invalid JSON schema: ...": Fix JSON syntax or schema structure according to JSON Schema standards."Authentication failed. Please check your API key.": Verify and update the API key credential."Rate limit exceeded or too many concurrent sessions. Please try again later.": Reduce request frequency or wait before retrying."Could not connect to Browser Use API. Please check if the service is available.": Check network connection and service status."Request to Browser Use API timed out. Please try again.": Retry the request; consider increasing timeout if persistent.