Gemini Search icon

Gemini Search

Perform searches and generate content using Google Gemini API

Overview

The Gemini Search node enables you to perform web searches using the Google Gemini API. Specifically, with the Web Search operation, it allows you to send a prompt to Gemini, optionally restrict the search to specific URLs, and receive AI-generated responses based on real-time web data. This is useful for scenarios where you need up-to-date information, summaries, or insights from the web within your n8n workflows.

Practical examples:

  • Summarizing recent news articles about a topic.
  • Extracting product information from specific websites.
  • Researching competitors by restricting search to their domains.

Properties

Display Name Type Description
Model options The Gemini model to use (e.g., Gemini 1.0 Pro, Gemini 2.0 Flash, etc.).
Prompt string The prompt to send to Gemini. Required.
Organization Context string Optional organization name to use as context for search.
Restrict Search to URLs string Optional comma-separated list of URLs to restrict search to (e.g., example.com,docs.example.com).
System Instruction string Optional system instruction to guide the model behavior.
Options collection Additional settings:
- Temperature number Controls randomness in the response (0-1). Default: 0.6
- Max Output Tokens number Maximum number of tokens to generate. Default: 2048
- Top P number Nucleus sampling parameter. Default: 1
- Top K number Top K sampling parameter. Default: 1
- Extract Source URL boolean Whether to extract the source URL from the response. Default: false

Output

The node outputs a JSON object with the following structure:

{
  "response": "<AI-generated text response>",
  "fullResponse": { /* Full Gemini API response object */ },
  "restrictedUrls": "<comma-separated URLs>", // Only present if 'Restrict Search to URLs' was set
  "sourceUrl": "<URL>",                       // Only present if 'Extract Source URL' is enabled and a source is found
  "redirectedSourceUrl": "<final URL after redirects>", // Only present if 'Extract Source URL' is enabled and redirection occurs
  "redirectError": "<error message>"          // Only present if there was an error resolving the redirected URL
}
  • response: The main text answer generated by Gemini.
  • fullResponse: The complete raw response from the Gemini API.
  • restrictedUrls: Echoes the restriction if provided.
  • sourceUrl: The original source URL extracted from the Gemini response (if enabled).
  • redirectedSourceUrl: The final destination URL after following HTTP redirects (if enabled).
  • redirectError: Error message if there was a problem resolving the redirected URL.

If an error occurs and "Continue On Fail" is enabled, the output will be:

{
  "error": "<error message>"
}

Dependencies

  • External Service: Requires access to the Google Gemini API.
  • Credentials: You must configure the geminiSearchApi credential in n8n.
  • Network Access: Outbound HTTPS requests to the Gemini API and any referenced URLs (for source extraction).

Troubleshooting

Common Issues:

  • Missing Credentials: If the geminiSearchApi credential is not configured, the node will fail.
  • Invalid Model Selection: Using an unsupported model value may result in errors from the Gemini API.
  • Prompt Required: The "Prompt" field is mandatory; leaving it empty will cause an error.
  • API Quotas/Rate Limits: Exceeding Gemini API quotas can lead to request failures.
  • Source URL Extraction Fails: If "Extract Source URL" is enabled but no source is found, sourceUrl will be empty. Redirect resolution may also fail due to network issues or timeouts, resulting in a redirectError.

Error Messages:

  • "Error fetching redirected URL: ...": Indicates a problem when trying to resolve the final URL after redirects. Usually caused by network issues or invalid URLs.
  • "Request failed with status code ...": Indicates an issue with the Gemini API request, possibly due to authentication, quota, or input errors.

How to resolve:

  • Double-check credentials and permissions.
  • Ensure required fields are filled.
  • Review API usage limits.
  • For source extraction issues, verify that the Gemini response includes grounding metadata.

Links and References

Discussion