MCP Server
The Scrayle MCP (Model Context Protocol) Server connects AI coding assistants like VS Code Copilot, Cursor, and Claude Desktop directly to the full Scrayle platform. Your AI agent can scrape pages, launch browsers, create and run workflows, manage storage, solve CAPTCHAs, and more - all through natural language.
What is MCP?
The Model Context Protocol is an open standard that lets AI assistants call external tools. When you connect the Scrayle MCP server to your editor, every Scrayle capability becomes a tool your AI can use directly in conversation.
Available Tools (33 total)
| Category | Tools | Description |
|---|---|---|
| Scraping | scrape_url, batch_scrape, crawl_website | Extract content from web pages as markdown, HTML, or structured data; recursively crawl entire sites |
| Browser Control | start_browser_session, browser_navigate, browser_click, browser_type, browser_screenshot, browser_get_content, browser_evaluate, browser_select, browser_scroll, browser_wait, close_browser_session, list_browser_profiles | Launch and directly control cloud browsers - navigate, click, type, screenshot, extract content |
| Workflows | create_workflow, list_workflows, run_workflow, get_workflow_status, list_workflow_runs | Create, list, execute, and monitor visual automation workflows |
| Marketplace | search_marketplace, get_marketplace_actor, run_marketplace_actor | Find and run pre-built scraping automations |
| CAPTCHA | solve_captcha | Solve reCAPTCHA, hCaptcha, and Turnstile challenges |
| Storage - KV | kv_create_namespace, kv_get, kv_set, kv_list_namespaces | Create and manage key-value store namespaces with get/set operations |
| Storage - Datasets | create_dataset, list_datasets, get_dataset_items, add_dataset_items | Create datasets and manage structured record collections |
How It Works
- The MCP server runs as an HTTP service at
https://mcp.scrayle.com/mcp - Your AI editor connects to it using the MCP protocol over HTTP with SSE streaming
- Authentication uses your Scrayle API key sent as an
x-api-keyheader - When you ask your AI to do something with Scrayle, it calls the appropriate tool
- Results are returned as formatted markdown directly in your conversation
💡Supported Editors
Any editor or application that supports MCP over HTTP (Streamable HTTP transport) can connect to Scrayle. This includes VS Code with GitHub Copilot, Cursor, Windsurf, and Claude Desktop.
Quick Links
- Setup Guide - Connect your editor to Scrayle MCP
- Tools Reference - Detailed reference for all 33 tools
- API Endpoints - REST API details for the MCP server