> ## Documentation Index
> Fetch the complete documentation index at: https://pathlit.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Browser Agent Node

> Automate web browsing tasks using an AI agent with Browserbase

## Overview

The Browser Agent Node provides intelligent web browser automation through Browserbase with Playwright and Stagehand tools. This node enables you to:

* Navigate websites and interact with web elements
* Extract data from web pages
* Perform complex browsing workflows
* Handle authentication and forms automatically
* Solve captchas and avoid detection

## Configuration Parameters

The Browser Agent uses Browserbase for remote browser sessions with advanced stealth and automation capabilities.

### Node Configuration

* **Task Instructions**:

  Describe the browsing task for the agent to accomplish.

  > *Search for Browserbase tutorials and summarize the first result*

  > *Login to the dashboard, navigate to settings, and extract the API key*

  > *Go to the pricing page and extract all plan details*

<AccordionGroup>
  <Accordion title="Advanced Settings">
    <ul>
      <li>
        <strong>Keep Session Running</strong>: Leave the browser session running after completion for reuse.
      </li>

      <li>
        <strong>Model</strong>: Select the language model for the agent.
      </li>

      <li>
        <strong>System Prompt</strong>: Customize the agent's browsing behavior.
      </li>

      <li>
        <strong>Max Steps</strong>: Maximum reasoning steps the agent can take (1-50).
      </li>

      <li>
        <strong>Enable Model Fallbacks</strong>: Allow automatic model fallbacks if needed.
      </li>
    </ul>
  </Accordion>

  <Accordion title="Browserbase Configuration">
    <ul>
      <li>
        <strong>Region</strong>: Preferred Browserbase region for new sessions.
      </li>

      <li>
        <strong>Session Timeout</strong>: Override session timeout in seconds (minimum 30s).
      </li>

      <li>
        <strong>Browser Context ID</strong>: Reuse a saved browser context.
      </li>

      <li>
        <strong>Viewport Size</strong>: Set viewport width and height in pixels.
      </li>

      <li>
        <strong>Log/Record Sessions</strong>: Enable session logging and rrweb recording.
      </li>

      <li>
        <strong>Solve Captchas</strong>: Toggle built-in captcha solver.
      </li>

      <li>
        <strong>Block Ads</strong>: Enable ad blocking.
      </li>

      <li>
        <strong>Advanced Stealth Mode</strong>: Control stealth protections.
      </li>

      <li>
        <strong>Proxy Options</strong>: Configure proxy routing and geolocation.
      </li>
    </ul>
  </Accordion>
</AccordionGroup>

## Expected Inputs and Outputs

* **Inputs**:

  * **input**: Optional templating inputs for the prompt
  * **session\_id**: Optional Browserbase session ID to reuse

* **Outputs**:

  * **output**: Agent's response summary
  * **conversation**: Full agent conversation log
  * **browser**: Browserbase session metadata (includes live view URL)

## Use Case Examples

1. **Web Data Extraction**:

   Use the agent to navigate complex websites and extract structured data without writing scraping code.

2. **Automated Testing**:

   Configure the agent to test web applications by performing user actions and verifying results.

3. **Form Automation**:

   Let the agent fill out and submit forms across multiple websites automatically.

4. **Competitive Intelligence**:

   Automate browsing competitor websites to gather pricing, features, and product information.

## Error Handling and Troubleshooting

* **Session Issues**:

  If the browser session fails, the node will handle cleanup automatically. Check Browserbase configuration if issues persist.

* **Element Not Found**:

  The agent uses natural language tools (Stagehand) that are more reliable than precise selectors. If precise tools fail, the agent will fall back to natural language.

* **Iframe Content**:

  When accessing iframe content, the agent will try to get the embedded page URL and navigate directly to it.

* **Page Crashes**:

  If the browser session crashes (ERR\_ABORTED), the agent will stop and report the error.

If you encounter any issues not covered in this documentation, please reach out to our [support team](mailto:dev@pathlit.ai) for assistance.

## Relevant Nodes

<CardGroup>
  <Card title="Web Extract Node" href="/docs/nodes/web-extract">
    Extract data from web pages
  </Card>

  <Card title="Web Search Node" href="/docs/nodes/web-search">
    Search the web for information
  </Card>

  <Card title="Agent Node" href="/docs/nodes/agent">
    General-purpose agent for complex tasks
  </Card>
</CardGroup>