What Scrapling's MCP actually does
D4Vinci's Scrapling MCP scrapes and drives pages from a local Python process. Install for Cursor and Claude, limits, and nearby listings on PlugMCP.
Scrapling is a Python scraping framework with an MCP server on top. The agent can fetch a page over HTTP, open a stealth browser session, click and fill forms, take screenshots, and pull Markdown or selected elements. It is local. Your machine runs the process.
This week it is moving hard. Appsgit's MCP trending list for 30 Oct to 6 Nov 2026 ranks Scrapling MCP Server near the top of the weekly star gainers. Treg's weekly GitHub trending crawl for the same window also surfaces D4Vinci/Scrapling. On PlugMCP the listing is under web scraping and browser automation.
What it is good for
Use it when the agent needs page text or structured fields more than a long interactive browser loop. One-shot tools cover a single HTTP request or a single stealth browser fetch. Session tools keep cookies and tabs across calls so the agent can walk a few product pages, save login state, or inspect network history without relaunching each time.
The MCP docs list seventeen tools. HTTP and stealth fetch, open and close sessions, screenshots and short observe sequences, extract as snapshot, HTML, Markdown or text, chained browser actions, JavaScript evaluate, and network request history. Ad and tracker domains are blocked in browser tools by default. Content sanitization aims to strip hidden prompt-injection material when main-content extraction is on.
The package on PyPI is scrapling (checked at 0.5.1 when this draft was written). The MCP entry point is scrapling-mcp, which maps to scrapling mcp.
Install in Cursor
Python 3.10 or newer. Install the AI extra, then the browser dependencies the fetchers need:
pip install "scrapling[ai]"
scrapling install
In Cursor, open Settings, then MCP, then add a server. Use a command of scrapling-mcp with no args for stdio, or paste this into your MCP config (point command at the full path from which scrapling-mcp if your client does not see the venv):
{
"mcpServers": {
"ScraplingServer": {
"command": "scrapling-mcp"
}
}
}
Docker is also documented:
{
"mcpServers": {
"ScraplingServer": {
"command": "docker",
"args": ["run", "-i", "--rm", "pyd4vinci/scrapling", "mcp"]
}
}
}
The PlugMCP listing currently shows uvx scrapling==0.5.1. The publisher docs and the 0.5.1 wheel use scrapling-mcp after pip install "scrapling[ai]". Prefer the docs path so the MCP extra and browsers are actually installed. After it connects, ask the agent to fetch a simple page as Markdown. You should see a tool result, not an empty tool list.
Install in Claude
Claude Code (use the path from which scrapling-mcp):
claude mcp add ScraplingServer "/path/to/scrapling-mcp"
Claude Desktop uses the same command scrapling-mcp block in claude_desktop_config.json, then fully quit and restart the app.
For Streamable HTTP instead of stdio, the docs say:
scrapling-mcp --http --auth-token "$(openssl rand -hex 32)"
Default listen is 127.0.0.1:8000. HTTP without a token refuses to start unless you pass --no-auth. Prefer the SCRAPLING_MCP_AUTH_TOKEN environment variable over putting the token on the command line.
Limits to know
- Local stdio. The agent starts a process. The PlugMCP listing is not a remote proxy.
- Close browser sessions when finished or they keep running.
- HTTP mode without auth exposes every tool to anyone who can reach the port.
- Respect
robots.txt, rate limits, and site terms. Bypass features are not a legal green light. - Cross-check the publisher MCP guide when the listing install block lags.
Alternatives on PlugMCP
- Firecrawl (firecrawl-firecrawl-mcp-server) when you want a remote scrape and search URL rather than a local Python process.
- Playwright MCP (microsoft-playwright-mcp) when the job is driving a persistent accessibility-tree browser more than scrape-and-extract.
- Chrome DevTools MCP (chromedevtools-chrome-devtools-mcp) when you care about performance traces, network detail, and Chrome-specific debugging.
- Apify (apify-mcp-server) when the work belongs in Apify Actors and a store of managed scrapers, with an API key.
On the PlugMCP page
The listing is d4vinci-scrapling. Transport is stdio. Status is local. The install block is the publisher's local command, not an mcp.plugmcp.ai proxy. Hosting a copy on PlugMCP is a separate action if you want keys and logs on a server you operate.
Search more scrape and browser servers from the directory MCP at https://plugmcp.ai/mcp-server, or open the web scraping and browser automation categories.
