Scrapling MCP Server

Content

Scrapling MCP Server brings Scrapling's scraping and browser-fetching capabilities into MCP clients. Claude can ask it to make fast HTTP requests, fetch dynamic content through Chromium or Chrome, use stealth fetching for protected pages, target specific elements with CSS selectors, capture screenshots from open sessions, and manage persistent browser sessions across requests.

The project publishes MCP metadata in server.json, documents a PyPI package runtime with scrapling mcp, and provides source and hosted documentation for the MCP server. Scrapling is especially relevant when an agent needs targeted extraction before handing page content to the model, which can reduce token use and avoid passing full pages when a selector is enough.

Source Review

• https://github.com/D4Vinci/Scrapling
• https://scrapling.readthedocs.io/en/latest/ai/mcp-server.html
• https://pypi.org/project/scrapling/
• https://github.com/D4Vinci/Scrapling/blob/main/README.md
• https://github.com/D4Vinci/Scrapling/blob/main/server.json
• https://github.com/D4Vinci/Scrapling/blob/main/docs/ai/mcp-server.md
• https://github.com/D4Vinci/Scrapling/blob/main/docs/api-reference/mcp-server.md
• https://github.com/D4Vinci/Scrapling/blob/main/pyproject.toml
• https://github.com/D4Vinci/Scrapling/blob/main/LICENSE

These sources were reviewed on 2026-06-06. Prefer the live repository, README, MCP metadata, source docs, API reference, package metadata, hosted docs, PyPI page, and license for current install commands, package versions, tool names, transport options, and scraping guidance.

Features

• Stdio MCP server through scrapling mcp.
• PyPI package metadata and MCP server.json package metadata for uvx usage.
• Optional Streamable HTTP mode through the Scrapling CLI.
• Basic HTTP scraping with get and bulk_get.
• Browser-backed dynamic fetching with fetch and bulk_fetch.
• Stealth browser fetching with stealthy_fetch and bulk_stealthy_fetch.
• Screenshot capture from an existing dynamic or stealth browser session.
• Persistent session management with open_session, close_session, and list_sessions.
• CSS selector targeting before content is handed to the model.
• Proxy support, browser impersonation, parallel processing, and session reuse.
• Automatic hidden-content sanitization when main_content_only is enabled.

Installation

Install Scrapling with MCP dependencies, then install browser dependencies before using browser-backed tools:

pip install "scrapling[ai]"
scrapling install

For Claude Code, add the stdio server after confirming the scrapling executable is on the PATH:

claude mcp add ScraplingServer scrapling mcp

For MCP clients that use JSON configuration:

{
  "mcpServers": {
    "ScraplingServer": {
      "command": "scrapling",
      "args": ["mcp"]
    }
  }
}

The project's MCP metadata also documents uvx with the fixed mcp package argument:

{
  "mcpServers": {
    "ScraplingServer": {
      "command": "uvx",
      "args": ["scrapling", "mcp"]
    }
  }
}

Restart the MCP client after adding the server.

Use Cases

• Extract a specific product field, table, article body, or listing element with a CSS selector.
• Use basic HTTP scraping for simple pages where browser automation is unnecessary.
• Fetch JavaScript-rendered pages through a browser-backed dynamic fetch.
• Use stealth fetching only for allowed workflows where normal fetching fails on protection pages.
• Capture screenshots from an existing session so Claude can inspect visual page state.
• Reuse a persistent session while scraping multiple pages from the same allowed site.
• Use bulk tools to scrape a small reviewed list of URLs more efficiently than repeated single requests.
• Keep hidden-content sanitization enabled to reduce prompt-injection exposure from scraped pages.

Safety and Privacy

Scrapling MCP is an automation-capable scraping server. Confirm permission, scope, rate limits, and legal basis before letting an agent scrape or screenshot a site. Its own documentation advises checking robots guidance, respecting rate limits, complying with site terms, and respecting copyright.

Use the least invasive tool that works. Prefer selector-targeted get requests for simple public pages, and reserve browser, stealth, proxy, screenshot, and bulk workflows for cases where they are allowed and necessary. Close persistent sessions when finished, and review scraped content for prompt injection, personal data, secrets, and copyrighted material before forwarding it into reports, datasets, or other tools.

Duplicate Check

Existing entries cover Browserbase, BrowserMCP, Bright Data, Chrome MCP, Playwright MCP, and related browser automation or scraping tools. No D4Vinci/Scrapling, Scrapling MCP, scrapling mcp, Scrapling selector scraping, or Scrapling stealth scraping entry was found in content/mcp, content/tools, content/guides, content/agents, or content/skills.