Overview
This node connects to a Gradio WebUI API endpoint and executes tasks with streaming support. It sends detailed configuration and task instructions to the API, then listens for streamed event data representing the progress and results of the task execution. This is useful for automating complex workflows that involve AI agents interacting with web browsers or other tools, where real-time feedback and stepwise execution details are important.
Typical use cases include:
- Automating web browsing tasks driven by natural language instructions.
- Running AI agents that perform multi-step actions with vision capabilities.
- Capturing live browser views, recordings, and agent thought/action traces during execution.
For example, you could instruct the node to "go to google.com, search for 'OpenAI', click the first result, and return the URL," while receiving live updates and final results streamed back from the Gradio WebUI service.
Properties
| Name | Meaning |
|---|---|
| API URL | URL of the Gradio WebUI API endpoint |
| Gradio API Key | Authentication key for accessing the Gradio WebUI API |
| Agent Type | Specifies the type of agent (e.g., "org") |
| LLM Provider | Large Language Model provider name (e.g., "gemini") |
| Model Name | Name of the LLM model to use (e.g., "gemini-2.0-flash-exp") |
| Temperature | Sampling temperature parameter for the LLM |
| Base URL | Base URL for the LLM provider's API (e.g., Google Gemini API endpoint) |
| LLM Provider API Key | API key for authenticating with the LLM service provider |
| Use Own Browser | Whether to use a dedicated browser instance |
| Keep Browser Open | Whether to keep the browser open after task completion |
| Headless Mode | Whether to run the browser in headless mode |
| Disable Security | Whether to disable security features in the browser |
| Window Width | Width dimension for the browser window |
| Window Height | Height dimension for the browser window |
| Recording Path | Filesystem path to save video recordings of the browser session |
| Agent History Save Path | Filesystem path to save the agent's action history |
| Trace Path | Filesystem path to save trace files related to the agent's execution |
| Enable Recording | Whether to enable recording of the browser session |
| Task Description | Natural language description of the task to perform |
| Additional Information | Extra information to provide context or instructions |
| Max Run Steps | Maximum number of steps the agent should execute |
| Use Vision | Whether to enable vision capabilities for the agent |
| Max Actions per Step | Maximum number of actions allowed per step |
| Tool Calling Method | Method used for calling external tools (e.g., "auto") |
Output
The node outputs a JSON object containing multiple fields extracted from the streamed events received from the Gradio WebUI API:
liveBrowserView: Live view data of the browser during task execution.finalResult: The final output/result of the executed task.errors: Any errors encountered during execution.modelActions: Actions taken by the AI model/agent.modelThoughts: Internal thoughts or reasoning steps of the AI model.latestRecording: Reference or path to the latest recorded browser session video.traceFile: Path or content of the trace file capturing execution details.agentHistory: History log of the agent's actions.
These fields may be null if not provided by the API response.
Dependencies
- Requires access to a Gradio WebUI API endpoint capable of handling the specified POST and GET requests.
- Needs valid API keys for both the Gradio WebUI API and the chosen LLM provider.
- Relies on n8n's HTTP request helper to send requests and receive streamed event data.
- The environment should allow network access to the configured API URLs.
Troubleshooting
- Request helper not available: The node depends on n8n's built-in HTTP request helper. Ensure your n8n version supports this feature.
- Authentication errors: Verify that the provided API keys are correct and have sufficient permissions.
- Streaming issues: If the node fails to parse streamed events, check the API endpoint stability and ensure it returns properly formatted Server-Sent Events (SSE).
- JSON parsing failures: Malformed or unexpected data in the stream can cause parsing errors; verify the API response format.
- Timeouts or incomplete data: Network interruptions or slow responses might lead to partial results; consider increasing timeouts or checking network reliability.