MrScraper for Dify

MrScraper is a native Dify Tool Plugin for web discovery, page rendering, AI extraction, scraper creation, scraper execution, and result retrieval. One provider-level API token enables 15 independent tools in Dify Agents, Chatflows, and Workflows.

Setup

Sign in to the MrScraper app and create or copy an API token.
Install this plugin in a compatible Dify instance.
Open the MrScraper provider credentials and enter the token under API Token.

The credential is validated with the MrScraper account-information endpoint. The token is supplied by Dify at runtime and is not a tool parameter.

Tools

Account

Get Account Info — account details, token use, and limits.

Discovery

Crawl Website URLs — immediately discover URLs from a starting website.
Search Google SERP — synchronous Google results as JSON or exact HTML.

Extraction

Extract Page by Prompt — immediately extract one page with optional JSON-shape instructions.
Extract Listings — immediately extract repeated or paginated items.
Extract Structured Data — use the bundled Article, Forum Thread, Hotel, Job Posting, Post, Product, Property, Restaurant, Social Media Profile, or Tour/Attraction preset.
Fetch Rendered HTML — render JavaScript in a stealth browser with optional Markdown, screenshot, cookies, selector waiting, geo/proxy settings, and home-page navigation.

Results

Get Results — paginated results with page size and sort order.
Get Latest Results — the newest N results.
Get Result Detail — one complete result by ID.

Scraper Creation

Create Prompt Scraper — create a reusable General AI scraper.
Create Listing Scraper — create a reusable Listing AI scraper.
Create Website Crawl Scraper — create a reusable Map AI scraper.

Scraper Runs

Run Existing Scraper AI - General/Listing/Map — run one URL with the settings supported by that AI agent type.
Run Existing Scraper Manual — run one URL with an existing manual scraper.
Run Existing Scraper Batch — run a nonempty array of URLs through the AI or manual bulk endpoint.

Examples

An Agent can call Extract Page by Prompt with:

For a Workflow, bind a previous node's array output to Run Existing Scraper Batch:

, , and accept native structured values. For compatibility with serialized Agent or Workflow values, the boundary also accepts one valid JSON string of the matching top-level type. Comma-splitting and Python-literal parsing are intentionally not used.

Outputs and timeouts

Upstream JSON objects and top-level arrays are emitted as Dify JSON messages without dropping fields. The pinned Dify Plugin SDK 0.10.2 supports top-level arrays directly, so no wrapper field is added. HTML and plain-text responses are emitted as exact text messages and are never mislabeled as JSON.

Scraping can be slow. The shared client uses a finite 10-second connection timeout and a 620-second default read timeout. Rendered-page calls use the configured page timeout plus a 30-second transport allowance. Dify, reverse proxies, and infrastructure may impose shorter limits.

Privacy, safety, and support

The plugin sends target URLs, prompts, selectors, output schemas, cookies, paginator settings, and related options to MrScraper. Review PRIVACY.md, the MrScraper privacy policy, and the MrScraper acceptable use policy.

Only scrape public content or content you are authorized to access. Follow the target site's terms and all applicable privacy, copyright, and computer-access laws.

Source and issue tracking: ai-mrscraper/dify-plugin-mrscraper. MrScraper support is available through mrscraper.com.