Jina AI n8n integrationAutomate Jina AI with n8n.
Pulling readable text out of a web page is usually the ugly part of an AI workflow. The Jina AI n8n integration takes care of it with 3 operations spread over 2 resources: read a URL, search the web, run a deep research report. No trigger node ships with it, so a Schedule Trigger or a webhook starts the run.
Verified Trustpilot reviews · AI, automation & growth agency
What does the Jina AI n8n integration actually do?
Jina AI turns web content into something a language model can read without choking on markup. The n8n node puts that behind 3 operations grouped in 2 resources: Reader, which holds Read and Search, and Research, which holds Deep Research. You drop the node into a workflow, pick an operation, fill the one required field, and the cleaned result moves down the canvas like any other item.
Start with the gesture, not the field list. Add a Schedule Trigger, add the Jina AI node, set the resource to Reader and the operation to Read, paste a page address in URL, hit Execute. What comes back is the page content in the format you chose, which you can hand straight to a model or write to Google Docs. That single chain is already a working scraper replacement, and everything else on this page is a variation on it.
Swap the operation to Search and the same node becomes a watch. You give it a Search Query, it returns the top results as clean content instead of raw HTML. Narrow it to the domains you care about with Site Filter, run it every morning, append what comes back as rows in Google Sheets. Nobody opens a browser tab for this again.
Deep Research is the heavier one. Instead of a keyword it takes a topic, and instead of links it returns a structured research report. A workflow that posts a supplier brief or a market note into a channel every Monday is three nodes long: trigger, Jina AI, delivery.
Two things are worth knowing before you build. All 3 operations carry the same Simplify switch, which returns a simplified version of the response instead of the raw data, so turn it on when a later node only needs the essentials. And whatever shape the result has, the next node reads it with an expression, the {{ $json.field }} syntax n8n uses to reach into the incoming item.
There is no Jina AI trigger in n8n, so nothing here reacts to an outside event on its own. A Schedule Trigger on a fixed interval, an n8n Webhook called by another service, or the trigger of a different tool gets the workflow moving. If you need an endpoint that the node does not cover, the HTTP Request node calls it directly and reuses the same credential through predefined authentication. Our n8n review goes into where that fallback is worth the trouble, and the n8n training covers building the chain end to end.
How do you connect Jina AI to n8n?
- 01
Grab your API key
Go to the Jina AI website and select API on the page. In the API app widget, select API KEY & BILLING. Copy the key labeled "This is your unique key. Store it securely!". That string is the whole authentication story here: the node supports one method, API key, and there is no authentication selector to fiddle with on the node itself.
- 02
Create the credential in n8n
Open the Credentials menu in n8n and create a Jina AI credential. A credential is the stored login n8n uses on your behalf, so the key never sits in the workflow itself. Paste the key into the API key field and save. You do this once: the same credential is then offered in the dropdown of every Jina AI node you add afterwards, in every workflow on the instance.
- 03
Run one operation to check it
Add the Jina AI node, pick the credential from its dropdown, set Resource to Reader and Operation to Read, then paste any public page in URL and execute the node alone. A required field left empty stops the node before it calls anything, so a clean run here means the key works and you can start wiring the rest around it.
What can the Jina AI node do?
The Jina AI node exposes 3 operations across 2 resources. For each one: the node as you configure it in n8n, the required fields, and our field notes.
Reader
2 operationsRead URL content
reader.readWhat you see in n8n
Notes & use cases
Give the node a page address and it returns the content already cleaned up, in the format you picked.
Key parameters
- URL: the page to fetch content from, typed in or taken from an earlier node with
{{ $json.field }}. - Output Format: HTML, JSON, Markdown, Screenshot or Text, 5 choices in the Options collection.
- Target CSS Selector and Exclude CSS Selector: the first focuses on specific page elements, the second drops what you never want, like a header or an ad block.
- Wait for CSS Selector: waits for a given element to appear before extracting, which is what dynamic pages need.
Search web
reader.searchWhat you see in n8n
Notes & use cases
Search runs a query through Jina AI and returns the top results as clean, readable content rather than a page of markup you then have to parse.
Key parameters
- Search Query: required, the words you would type into a search box, or an expression such as
{{ $json.field }}to make each run different. - Site Filter: restricts the search to specific websites, useful when only two or three domains matter to you.
- Page Number: which page of the search results to retrieve, when the first one is not enough.
Research
1 operationPerform deep research
research.deepResearchWhat you see in n8n
Notes & use cases
This one takes a topic rather than a keyword and generates a structured research report on it.
Key parameters
- Research Query: required, the topic or question the AI researches, written as a full question rather than two words.
- Max Returned Sources: the maximum number of URLs included in the final answer.
- Prioritize Sources and Exclude Sources: two domain lists, the first given higher priority for content retrieval, the second strictly excluded from it.
Need help automating Jina AI with n8n?
A person reads every message.