Zenrows

Author: zenrows Version: 0.2.1 Type: tool Repository: https://github.com/ZenRows/dify-zenrows Contact: support@zenrows.com

Description

Fetch and extract data from any website with Zenrows, from inside your Dify workflows and agents. Zenrows handles JavaScript rendering, anti-bot protection and proxy rotation, so a tool call returns the page content rather than a block page.

Tools

Tool	What it does
Fetch	Retrieve one page. Optional JavaScript rendering, premium proxies, country geolocation, wait-for-selector and full-page screenshot. Returns HTML, Markdown, plain text, PDF or an image.
Extract	Return structured JSON instead of page content — site-tailored extraction, general-purpose autoparse, your own CSS selectors, or built-in filters for emails, links, tables and similar.
Batch Create	Submit up to 1000 URLs as a single asynchronous job. Returns a job ID, or waits for the job if it is small.
Batch Status	Check how far a job has progressed, how many tasks failed, and what it has cost so far.
Batch Results	Collect the scraped content from a finished job.

Use Fetch when you want the page. Use Extract when you want fields out of the page. Use Batch when you have more URLs than one request should carry.

Setup

1. Get a Zenrows API key from https://app.zenrows.com — the dashboard shows it under your account.
2. In Dify, go to Integrations → Tool Plugin, install Zenrows, and paste the key when prompted.
3. Add a Zenrows tool to any workflow or agent.

Connection requirements. The plugin makes outbound HTTPS requests to api.zenrows.com and async.api.zenrows.com. Both must be reachable from wherever your Dify instance runs. No inbound connections and no local services are required.

Usage

Add a Zenrows node to a workflow and pass it a URL. The tool returns the page content, which you can feed into an LLM node, a knowledge base, or any downstream step.

Changes in 0.2.1

Fixed: the named output variables of Fetch (content, url, status_code) and Extract (data, method_used, fell_back_to_autoparse, empty, url) now reach later workflow steps. In 0.2.0 they were empty, and only the text and json outputs carried the result.

Changes in 0.2.0

Behavior change: Batch Create now uses Adaptive stealth by default. Like Fetch and Extract, Batch Create lets Zenrows escalate each URL to JavaScript rendering or premium proxies when its site needs them. Protected sites that failed on 0.1.0 now succeed, but escalated URLs cost more credits. To keep the 0.1.0 behavior, turn Adaptive stealth off on the Batch Create node.

Fixed:

• Fetch with a proxy country no longer fails with REQS004 under Adaptive stealth.
• Extract falls back to autoparse on a domain not prepared for Extract yet, instead of reporting that your API key was rejected.

Development

Working on the plugin itself, or reviewing it? See CONTRIBUTING.md — running it against Dify in debug mode, building the .difypkg, validating it, installing it locally, and the known gotchas.

License

MIT — see LICENSE.