Package Information
Available Nodes
Documentation generating…
Documentation
n8n-nodes-oxylabs-web-api
This is an n8n community node for the Oxylabs Web API. It lets your workflows and AI agents search the live web and read any page, including JavaScript-heavy and bot-protected ones, as Markdown, HTML, a screenshot or structured JSON.
n8n is a fair-code licensed workflow automation platform.
Installation ·
Operations ·
Credentials ·
Compatibility ·
Usage ·
Resources
Installation
Follow the installation guide in the n8n community nodes documentation, and install the package n8n-nodes-oxylabs-web-api.
Operations
| Resource | Operation | What it does |
|---|---|---|
| Search | Search the Web | Ranked organic results (title, outline, URL) for a query, optionally for a country |
| Scrape | Scrape a URL | Read one page as Markdown (default), HTML, a PNG screenshot (binary data), or the fields you describe as JSON |
| Scraper | Run | Run a dedicated scraper (Amazon, Google, Walmart, ChatGPT, YouTube and many more) with its own parameters, as content or parsed JSON |
| Scraper | Get Parameters | The JSON Schema of the parameters a scraper accepts |
| Request | Get Result | The result of a queued scrape, by its request_id |
The node can also be used as a tool by the n8n AI Agent.
Credentials
- In the Oxylabs dashboard, create a Web API instance, open it, and copy the API key shown at the top of the instance page.
- In n8n, create an Oxylabs Web API credential and paste the key.
The key is sent as Authorization: Bearer <key>. Test checks it with a free call that lists the scrapers.
Compatibility
Tested with n8n 2.x (local npm run dev). Requires n8n with n8nNodesApiVersion 1.
Usage
Examples
Research agent. Add an AI Agent node with a chat model, and attach Oxylabs Web API as a tool twice: once with Search the Web, once with Scrape a URL. Click the sparkle next to Query and URL to let the model fill them in. The agent can then search for sources and read the pages it picks.
Read a list of pages. Feed rows with a url column (for example from Google Sheets) into Scrape a URL, set URL to the expression {{ $json.url }} and Format to Markdown. Each row comes out with the page in its markdown field, ready for an LLM or a document store.
Pull fields out of a page. In Scrape a URL, choose Structured Data (JSON) and describe the fields, for example the book titles and their prices. The item's json field holds the result.
Track Google results. Start with a Schedule Trigger, then Scraper > Run with scraper scrape/google/search, Parameters {"query": "best running shoes", "location": "US"} and Output Structured Data (JSON). Each run returns the parsed results page, which you can append to a sheet to watch rankings over time.
Slow pages and queued scrapes
Every scrape is tried synchronously first. If it outlives the synchronous call, the node sends the same request to the API's queue and waits for it, checking every 5 seconds, for up to Max Wait (5 minutes by default). Scrapers marked (queued only) are always queued.
If the wait runs out, or Wait for Queued Results is off, the item carries a request_id and state: "pending". Fetch it later with Request > Get Result.
Structured data
- Scrape a URL with the Structured Data (JSON) format uses the AI parser. Describe the fields in plain words, or give a JSON Schema.
- Scraper > Run with Structured Data (JSON) uses the scraper's dedicated parser, which returns search results, products, sellers, hotels and the like in a fixed shape. Set AI Fallback Prompt to have the AI parser step in when the dedicated parser fails, returns nothing, or does not exist for that scraper. Choose the AI parser to skip the dedicated one.
Each output item says which parser answered in its parser field.
Scraper parameters
Parameters is the scraper's own request body as JSON, for example {"query": "wireless earbuds", "domain": "de"} for scrape/amazon/search. Run Scraper > Get Parameters to see every key, its type and its allowed values.
Errors
A wrong key, a spent quota (429 QUOTA_EXCEEDED) and invalid parameters fail at once with a message that says what to fix. Rate limits and transient server errors are retried twice with backoff. Faulted requests are not charged; quote the request_id in the error to Oxylabs support if one keeps failing.
Resources
- n8n community nodes documentation
- Oxylabs Web API remote MCP server, for agents that speak MCP: n8n's MCP Client Tool can connect to
https://mcp.oxylabs.io/web-api-mcpwith the same key