🇬🇧 EN

🌐 Web Search

Performs a live web search and returns a list of results including titles, URLs, and snippets. Useful for giving AI agents access to up-to-date information, news, or research from the public web.

Category: Data Retrieval & Processing · Type identifier: web_search

Overview

Large language models have a knowledge cutoff date and cannot access information published after their training. The Web Search node bridges this gap by querying the live web and returning fresh results that can be passed as context to an Agent node. This is the foundation of a search-augmented generation (SAG) pattern.

The node issues a search query to the configured provider and returns a structured list of results, each containing the page title, URL, a short snippet, and the source domain. You can control how many results to return and whether to filter adult content.

To use this node, configure a search provider API key in Settings → Integrations. Supported providers include SerpApi, Brave Search, and Bing. Each provider requires its own API key.

Configuration

Field	Status	Description
Query	Required	The search query string. Supports `{{ variable }}` references.
Provider	Required	The search provider to use (e.g. SerpApi, Brave Search, Bing).
Limit	Optional	Maximum number of results to return. Default: 10. Maximum: 50.
Safe Search	Optional	Toggle. When enabled, filters adult content from results.

Output Data

The node returns an array of search result objects, one per result returned by the provider.

Field	Type	Description
`results`	array	Array of search result objects
`results[*].title`	string	Page title
`results[*].url`	string	Full URL of the result
`results[*].snippet`	string	Short excerpt from the page
`results[*].source`	string	Domain name of the result
`count`	number	Number of results returned

{{ web_search.output.results }} {{ web_search.output.results[0].title }} {{ web_search.output.results[0].url }} {{ web_search.output.results[0].snippet }} {{ web_search.output.count }}

Example Usage

Schedule trigger → Web Search for "latest AI news" → Agent summarises results.

Add a Schedule Trigger node set to run daily at 08:00.
Add a Web Search node. Set Query to latest AI news today, Provider to Brave Search, and Limit to 10. Use the step key web_search.
Add an Agent node. Pass {{ web_search.output.results }} as context and instruct the agent to write a concise daily briefing.
Add an Email Notification node to deliver the briefing to your inbox.

Tips & Notes

Snippets are short excerpts, not full page content. If you need the full text of a result page, use the HTTP Request node to fetch the URL returned in results[*].url.
Use a Reranker after Web Search for better precision. If you are passing results to an Agent, consider adding a Reranker node between the Web Search and Agent nodes to surface the most relevant results first.
Dynamic queries are supported. The Query field supports {{ variable }} references, so you can build queries from earlier step output, trigger data, or user input.
Each provider has different result quality and cost. Brave Search offers privacy-focused results; SerpApi mirrors Google results; Bing offers Microsoft's index. Test with your use case to find the best fit.
Safe Search is off by default. Enable it for workflows that may be exposed to end users or that process queries from untrusted sources.

Related Nodes

Reranker — re-rank Web Search results by relevance before passing them to an agent.
RAG Query — search your own knowledge base instead of the public web.
HTTP Request — fetch the full content of a URL returned by Web Search.
Agent — consume search results as context for AI-generated responses.