Haunt API — Web Extraction for AI Agents

hauntapi.com · Docs · Get API Key

A hosted MCP (Model Context Protocol) server that gives AI agents structured web extraction. Pass a URL and a plain-English prompt — get back clean JSON. No browser management, no parsing, no infrastructure.

Install

claude plugin add github:Darko893/haunt-mcp-server

Or install manually via Claude Code's plugin discover: /plugin > Discover

What it does

Two tools, zero setup:

Tool	Description
web_extract	Extract structured data from any URL. Provide a URL and a description of what you want.
get_usage	Check your current API usage and remaining credits.

Handles JavaScript rendering, Cloudflare protection, and complex page layouts automatically.

Setup

1. Get a free API key at hauntapi.com/#signup (100 requests/month free)
2. Install the plugin: claude plugin add github:Darko893/haunt-mcp-server
3. Pass your API key as the api_key parameter on each tool call

No environment variables, no config files, no local server to run.

Example usage

Ask Claude to extract data from any URL:

Extract the product name, price, and availability from https://example.com/product

Claude will call web_extract with:

• url: the target page
• prompt: "Extract the product name, price, and availability"
• api_key: your API key

You get back structured JSON with exactly the fields you asked for.

Pricing

Plan	Requests	Price
Basic	100/mo	Free
Pro	Pay per request	$0.01/req
Business	5,000/mo	$49/mo
Enterprise	25,000/mo	$99/mo

Details at hauntapi.com.

Manual MCP config (without plugin)

If you prefer to configure manually instead of using the plugin:

Claude Desktop (SSE transport):

{
  "mcpServers": {
    "haunt": {
      "url": "https://hauntapi.com/sse"
    }
  }
}

Cursor / VS Code / Windsurf (Streamable HTTP):

{
  "mcpServers": {
    "haunt": {
      "url": "https://hauntapi.com/mcp/"
    }
  }
}

Plugin structure

haunt-mcp-server/
├── .claude-plugin/
│   └── plugin.json        # Plugin manifest
├── .mcp.json              # MCP server configuration
├── skills/
│   └── web-extraction/
│       └── SKILL.md       # Extraction best practices
├── README.md
├── METADATA.md
└── package.json           # Smithery/Glama metadata

Links

• Website: hauntapi.com
• Documentation: hauntapi.com/docs
• API Status: hauntapi.com/health
• MCP Status: hauntapi.com/mcp-health

License

This is a hosted commercial service. The plugin configuration and documentation are provided for integration purposes. The extraction engine itself is proprietary. See hauntapi.com for terms of service.