Skip to main content

MCP Server Integration

Learn how to connect AI assistants like Cursor and automation platforms like n8n to DumplingAI using the Model Context Protocol (MCP).

What is MCP?

Model Context Protocol (MCP) is an open standard that enables AI assistants and automation tools to connect to external data sources and APIs. With DumplingAI’s MCP server, you can access all DumplingAI tools directly from your AI assistant or workflow automation platform.

Benefits

  • Direct tool access: Use DumplingAI tools without writing API integration code
  • AI assistant integration: Connect Cursor, Windsurf, and other MCP-compatible clients
  • Workflow automation: Build n8n workflows that leverage DumplingAI’s capabilities
  • Simplified development: Let AI assistants handle API calls and data processing

Prerequisites

Before you begin, make sure you have:
  • A DumplingAI account with an API key (get one here)
  • An MCP-compatible client (Cursor, Windsurf, or n8n)

Available Tools

DumplingAI’s MCP server provides access to tools across multiple categories:

Web Scraping & Data Extraction

  • scrape - Fetch structured data and HTML from URLs
  • crawl - Recursively scrape multiple pages
  • screenshot - Capture webpage screenshots
  • extract - Extract structured data using custom schemas

Search & Discovery

  • search - Perform live web searches with previews
  • search_youtube - Search YouTube videos
  • search_maps - Search Google Maps locations
  • search_places - Find places and businesses
  • search_news - Search news articles
  • search_tiktok_users - Find TikTok users

AI & Processing

  • generate_agent_completion - Build and interact with AI agents
  • generate_ai_image - Generate images using AI

Social Media & Video Data

  • YouTube: channel info, videos, shorts, transcripts, comments
  • TikTok: profiles, videos, transcripts, comments, followers
  • LinkedIn: profiles, company data

Document Processing

  • doc_to_text - Extract text from documents
  • convert_to_pdf - Convert documents to PDF
  • merge_pdfs - Combine PDF files
  • trim_video - Trim video files
  • extract_document - Extract document metadata
  • read_pdf_metadata / write_pdf_metadata - Manage PDF metadata

Developer Tools

  • run_js_code - Execute JavaScript code
  • run_python_code - Execute Python code

Quick Start

Choose your preferred connection method:
MethodBest ForSetup Complexity
Remote (Streamable HTTP)Cursor, Windsurf, n8n⭐ Easy - no installation
Local (stdio)Claude Desktop, legacy clients⭐⭐ Moderate - npm package
Best for: Cursor, Windsurf, n8n, and other clients that support HTTP Streamable transport. Advantages:
  • No installation required
  • Always up-to-date with latest features
  • Simpler configuration
  • Works across different machines
How it works: Your client connects directly to https://app.dumplingai.com/api/mcp using HTTP Streamable transport with Bearer token authentication.

Cursor

Cursor supports MCP servers directly via HTTP Streamable transport.
  1. Open your Cursor MCP configuration file at ~/.cursor/mcp.json
  2. Add the DumplingAI MCP server:
{
  "mcpServers": {
    "DumplingAI": {
      "url": "https://app.dumplingai.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
  1. Replace YOUR_API_KEY with your DumplingAI API key
  2. Restart Cursor
  3. Verify the connection: The DumplingAI tools should appear in Cursor’s MCP tools list

Windsurf

Windsurf also supports MCP servers.
  1. Open your Windsurf MCP configuration file at ~/.codeium/windsurf/mcp_config.json
  2. Add the DumplingAI MCP server:
{
  "mcpServers": {
    "DumplingAI": {
      "url": "https://app.dumplingai.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
  1. Replace YOUR_API_KEY with your DumplingAI API key
  2. Restart Windsurf
  3. Verify the connection: Check that DumplingAI tools appear in Windsurf’s tools panel

n8n Workflow Automation

n8n can use DumplingAI tools in AI agent workflows via the MCP Client Tool node.
  1. In your n8n workflow, add an AI Agent node
  2. Add an MCP Client Tool node
  3. Create new MCP credentials:
    • Connection Type: HTTP Streamable
    • URL: https://app.dumplingai.com/api/mcp
    • Authentication: Header Auth
      • Header Name: Authorization
      • Header Value: Bearer YOUR_API_KEY
  4. Connect the MCP Client Tool to your AI Agent
  5. The DumplingAI tools will be available to your AI agent in the workflow
Make sure to set the N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE environment variable to true if using community packages.

Local MCP Server

Best for: Claude Desktop and clients that require stdio-based connections. Advantages:
  • Works with Claude Desktop
  • Compatible with older MCP clients
  • Runs locally on your machine
How it works: The npm package mcp-server-dumplingai runs locally and connects to DumplingAI’s API on behalf of your client using stdio transport.

Installation

Install the DumplingAI MCP server package:
npx -y @smithery/cli install @Dumpling-AI/mcp-server-dumplingai --client claude

Configuration for Claude Desktop

  1. Install via Smithery (recommended):
npx -y @smithery/cli install @Dumpling-AI/mcp-server-dumplingai --client claude
  1. Or manually configure in ~/Library/Application Support/Claude/claude_desktop_config.json:
{
  "mcpServers": {
    "dumplingai": {
      "command": "npx",
      "args": ["-y", "mcp-server-dumplingai"],
      "env": {
        "DUMPLING_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}
  1. Replace YOUR_API_KEY with your DumplingAI API key
  2. Restart Claude Desktop
  3. Verify the connection: Look for the 🔨 hammer icon in Claude Desktop indicating tools are available

Configuration for Other Clients

For clients that support stdio-based MCP (like older versions of Cursor or Windsurf):
{
  "mcpServers": {
    "dumplingai": {
      "command": "npx",
      "args": ["-y", "mcp-server-dumplingai"],
      "env": {
        "DUMPLING_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}
The local MCP server is maintained in a separate repository: github.com/DumplingAI/mcp-server-dumplingai

Example Use Cases

Web Research with AI Assistant

Ask your AI assistant in Cursor:
"Use the search tool to find recent news about AI developments,
then scrape the top 3 articles and summarize them"
The AI will:
  1. Use the search tool to find relevant articles
  2. Use the scrape tool to extract content from each article
  3. Summarize the findings

Content Extraction Workflow in n8n

Build an n8n workflow that:
  1. Receives a YouTube video URL
  2. Uses get_youtube_video to fetch video details
  3. Uses get_youtube_transcript to get the transcript
  4. Uses generate_agent_completion to summarize the content
  5. Saves the summary to your database

Social Media Research

Ask your AI assistant:
"Find the top 5 TikTok creators in the fitness niche,
get their latest videos, and analyze their content strategy"
The AI will use multiple tools:
  • search_tiktok_users
  • get_tiktok_profile_videos
  • get_tiktok_video

Best Practices

  1. API Key Security: Never commit your API key to version control. Use environment variables or secure credential storage.
  2. Credit Management: Monitor your credit usage in the DumplingAI dashboard. Each tool call consumes credits based on the operation.
  3. Rate Limits: Respect rate limits based on your subscription plan. Higher tiers offer increased rate limits.
  4. Tool Selection: Use the most specific tool for your needs. For example, use search_youtube instead of generic search for YouTube content.
  5. Error Handling: In n8n workflows, implement error handling nodes to manage API failures gracefully.
  6. Local Development: For local testing, use http://localhost:3000/api/mcp as your MCP URL when running the DumplingAI development server.

Comparison: Remote vs Local

FeatureRemote (Streamable HTTP)Local (stdio)
InstallationNone requirednpm package
SetupAdd URL + API keyInstall package + configure
MaintenanceAuto-updatedManual updates
Best forCursor, Windsurf, n8nClaude Desktop
AuthenticationBearer token in headersEnvironment variable
PerformanceDirect HTTPS connectionLocal process

Troubleshooting

Remote MCP Server Issues

Problem: MCP server not connecting Solutions:
  • Verify your API key is correct and active in the dashboard
  • Check that the URL is exactly https://app.dumplingai.com/api/mcp (no trailing slash)
  • Ensure the Authorization header format is Bearer YOUR_API_KEY
  • Restart your MCP client after configuration changes
  • Check client logs for specific error messages

Local MCP Server Issues

Problem: Package not found or installation fails Solutions:
  • Ensure you have Node.js 18+ installed
  • Try clearing npm cache: npm cache clean --force
  • Install globally: npm install -g mcp-server-dumplingai
  • Check that DUMPLING_API_KEY environment variable is set correctly
Problem: Tools not appearing in Claude Desktop Solutions:
  • Verify the config file location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)
  • Check that the JSON syntax is valid
  • Ensure Claude Desktop developer mode is enabled in Settings
  • Restart Claude Desktop completely

General Issues

Problem: Rate limit errors (429) Solutions:
  • Check your current usage in the DumplingAI dashboard
  • Upgrade your plan for higher rate limits
  • Implement delays between requests in automated workflows
Problem: Not enough credits Solutions:
  • Check your credit balance in the dashboard
  • Upgrade to a plan with more credits
  • Enable auto-recharge in your account settings
Problem: Tool returns error Solutions:
  • Verify the tool parameters match the documentation
  • Check that required parameters are provided
  • Ensure URLs and inputs are properly formatted
  • Review error messages for specific guidance

Next Steps

Support

Need help? Contact us at help@dumplingai.com
I