Build a research agent with n8n

Install

Runs on n8n Cloud, or on your own Docker instance. No packages to install.

Already have n8n wired up? Then8n integration pagehas the connection config on its own — this page assumes you have it and gets on with building.

Step by step

1. 1

   Store the key as a credential

   Create a Header Auth credential with name `x-api-key` and your key as the value. Never paste the key into a node parameter — credentials are encrypted and are not included when you export a workflow.

2. 2

   Add an HTTP Request tool

   Add an AI Agent node, then attach an HTTP Request Tool to it. The description you write is what the model reads when deciding whether to call it, so it matters more than it looks.

   {
     "name": "unlob_search",
     "description": "Search the web on unlob's own index. Returns metadata only — url, title, snippet and independent_sources. Never returns page bodies. Set min_independent_sources to 3 for factual claims.",
     "method": "GET",
     "url": "https://api.unlob.com/search",
     "authentication": "genericCredentialType",
     "genericAuthType": "httpHeaderAuth",
     "sendQuery": true,
     "queryParameters": {
       "parameters": [
         { "name": "q", "value": "={{ $fromAI('query', 'What to search for', 'string') }}" },
         { "name": "min_independent_sources", "value": "={{ $fromAI('min_sources', 'Independent sources required', 'number', 2) }}" },
         { "name": "limit", "value": "8" }
       ]
     }
   }

3. 3

   Schedule it

   Put a Schedule Trigger in front and a Slack or email node behind, and you have a monitoring workflow that runs itself. This is the shape n8n is genuinely best at.

What happens when it runs

What will go wrong

The failures that cost an afternoon rather than a minute, because they produce something that looks like it is working.

• !

  Use a Header Auth credential rather than typing the key into the URL or a parameter. Credentials are encrypted and are excluded from workflow exports — a key pasted into a node travels with the JSON when you share it.

• !

  `$fromAI()` is what lets the model fill a parameter. A hard-coded query value produces a tool that runs the same search every time and an agent that appears to ignore the question.

• !

  The tool description is the only thing the model sees. "Search the web" gets you a tool that is called for everything; describing what it returns and when to use it is what makes the agent selective.

• !

  Pin `min_independent_sources` in the node rather than exposing it to the model when you want a guaranteed corroboration floor. Anything left to $fromAI is a suggestion the model can lower.