How to Build an n8n Web Search Node Workflow With the You.com APIs

TLDR: The You.com n8n node brings six operations into your n8n workflows: web search, content extraction, answer synthesis, research, finance research, and background task management. It also works as an AI agent tool, so n8n's built-in agents can call You.com search on demand. This guide walks the install, the node's parameters, and a working agent-tool pattern, with every fact sourced from the You.com docs and the node's own README.

What is the n8n web search node? In n8n terms, a web search node is any node that issues a web search from inside a workflow. The You.com n8n integration is one such node: the @youdotcom-oss/n8n-nodes-youdotcom package provides a single node with six operations, Search, Get Contents, Answer, Research, Finance Research, and Get Research Task, built on the You.com Web Search and Contents APIs.

n8n is a fair-code licensed workflow automation platform where you connect APIs, build agents, and automate tasks in a visual editor. If you are building anything in n8n that needs fresh web data, a search node is usually the first thing you add, because most workflows that touch the outside world start with a query.

How Do You Install the You.com n8n Node?

Installation follows the standard community node path, and setup is two steps: install the package, then paste a credential.

Install the node by following the installation guide in the n8n community nodes documentation. In a self-hosted n8n instance, that means Settings, then Community Nodes, then Install, and entering @youdotcom-oss/n8n-nodes-youdotcom, the package published on GitHub and npm. On n8n Cloud, community nodes are managed by the platform.

Then set up credentials: get an API key from you.com/platform, open Credentials in n8n, create a new You.com API credential, paste the key, and click Test to verify it works. That credential is shared by all six operations of the node.

What Operations Does the Node Expose?

Six operations cover the search-to-research surface, each documented on the You.com n8n integration page and in the node README (fetched 2026-09-09).

Search searches the web and news with the You.com Web Search API. It sends a POST request with a JSON body and returns structured results with titles, URLs, descriptions, and snippets. Required: the query. Optional parameters include count (max results per section, 1 to 100, default 10), country, freshness (day, week, month, year, or a date range), language, safesearch, offset for pagination, and domain include, exclude, and boost lists of up to 500 domains each.

Get Contents extracts clean, structured content from one or more URLs you specify, returning page text as markdown or HTML plus metadata like JSON-LD, OpenGraph, and Twitter Cards. This is the operation for reading a specific page you already have the URL for.

Answer returns a synthesized answer with inline citations from web search results, for factual questions that need a quick, sourced answer.

Research returns a comprehensive, cited answer to a complex question: the Research API searches the web, reads multiple sources, and synthesizes a detailed markdown response with numbered citations. Effort levels run from lite through standard (the default), deep, exhaustive, and frontier.

Finance Research gives finance-grade research answers with citations, built on a retrieval index optimized for earnings reports, SEC filings, analyst coverage, and financial news.

Get Research Task polls the status of a background research task queued from the Research operation's Background mode. The README's recommended pattern is to put this behind a Wait node and loop until the status is no longer pending, because a poll that comes back still running just costs another loop iteration and nothing is lost to a timeout.

How Do You Build a Search Workflow With the Node?

The simplest useful workflow is a scheduled search that lands results in a spreadsheet or Slack. Add a Schedule Trigger, add the You.com node, choose the Search operation, set the query, and wire the output into any downstream node. The node returns structured JSON, so downstream nodes read fields directly, with no HTML parsing.

The parameter surface maps one-to-one onto the Web Search API. A monitoring query that only trusts certain sources uses the domain filters: include domains for a strict allowlist, exclude domains to filter sources out, boost domains to prefer some sources without excluding others. One constraint from the docs matters at execution time: include domains cannot combine with exclude or boost domains, and that pairing returns a 422. Exclude and boost can be combined freely.

For fresh results, the freshness parameter filters by recency: day, week, month, or year. A daily news digest workflow sets freshness to day and runs once per morning. A competitive-intelligence workflow might run weekly with freshness set to week and an exclude list for the company's own domain.

Search operators also work inside the query string itself: site: to restrict to a domain, filetype: to filter by file type, plus and minus terms, and boolean AND, OR, NOT. That gives you a second filtering layer without touching the parameter panel.

How Do You Use the Node as an AI Agent Tool?

The node has usableAsTool enabled, which means n8n's built-in AI agents can call it directly. Add the You.com node as a tool in any AI Agent workflow, and the agent can search the web or extract page content on its own, deciding when to query based on the task you give it.

The pattern that works: an AI Agent node with the You.com node attached as a tool, plus a chat trigger. The agent decides when to search, reads the structured results, and answers with sources. Because the results come back with URLs and snippets as fields, the agent can cite what it found without you writing any parsing code.

A concrete failure mode to watch for in agent workflows: the agent calls the search tool on every turn, even when the answer is already in the conversation. Detect it in n8n's execution log by counting tool calls per run. The fix is prompt-level: instruct the agent to search only when the task requires information it does not already have.

What Does the Attribution Header Do?

Every API request from the node includes an X-Client-Info attribution header that identifies the plugin automatically, with no configuration needed. The header identifies the plugin and its version, using the same convention the LangChain integration uses. You do not need to set anything for this to work, and it is how traffic from your n8n workflows is distinguishable from traffic from your other integrations.

When Should You Use the Node Instead of a Direct API Call?

The tradeoff is visual composition against code control. Use the node when the workflow's value is the chain: trigger, search, transform, notify, all visible in one canvas that a nontechnical teammate can inspect. Use the You.com Python or TypeScript SDK directly when you need the full parameter surface in code, custom error handling, or the workflow to live inside your application rather than beside it.

The two are not mutually exclusive. A common shape is n8n for the operational workflows (digests, alerts, monitoring) and direct SDK calls inside the product for user-facing search. Both authenticate with the same API key from you.com/platform.

For the direct-call path, the Python guide for the Web Search API covers the SDK with typed responses and retry configuration, and the Claude Code web search tool guide shows the same retrieval layer wired into a coding agent.

Where Does This Fit With the Rest of the Stack?

The node is one piece of the You.com integration surface. For agents outside n8n, the MCP server comparison guide covers connecting Claude, Cursor, and other MCP clients to You.com search. For evaluating whether the retrieval layer is good enough for your workflows before you build on it, the web search API evaluation guide covers the harness pattern that isolates retrieval quality as the only variable.

Next action: install the node in a scratch n8n instance, run one Search operation with a query from your actual workload, and confirm the result fields your downstream nodes would read. Then attach it to an AI Agent as a tool and watch the execution log to see when the agent chooses to search.