Which integrations are available for this server?

Extract canonical URL metadata from web pages as part of the metadata extraction capabilities, identifying the preferred URL for content. Automatically bypass Cloudflare protection when extracting data from URLs, handling JavaScript rendering and Cloudflare security measures transparently. Access the MCP server source code and documentation on GitHub, with the server hosted at the hauntapi/mcp-server repository. Handle JavaScript-rendered content when extracting data from URLs, allowing extraction from modern web applications and dynamic websites. Extract Open Graph metadata (Facebook/Meta tags) and other social media metadata from web pages as part of the metadata extraction capabilities.

Haunt API MCP Server

This is the canonical public MCP package working tree for @hauntapi/mcp-server and GitHub Darko893/mcp-server. The live Haunt API app/site lives separately in /root/haunt. Do not confuse this with /root/haunt/mcp-server, which is a legacy/local copy inside the app repo.

Give Claude, Cursor, Windsurf, and other MCP-compatible agents a clean web extraction tool.

Haunt turns messy public web pages into structured JSON using natural-language prompts. It is built for agent workflows that need product data, competitor pricing, article content, metadata, lead lists, research snippets, or any other web data without maintaining brittle selectors.

Quick Start

Get a free API key: https://hauntapi.com/#signup

Add this to your MCP client config:

{
  "mcpServers": {
    "haunt": {
      "command": "npx",
      "args": ["-y", "@hauntapi/mcp-server"],
      "env": {
        "HAUNT_API_KEY": "your-api-key"
      }
    }
  }
}

Then ask your agent:

Use Haunt to extract the product name, price, availability, and review count from https://example.com/product

Tools

`extract_url`

General-purpose extraction from any web page.

Use it for:

Product names, prices, stock status, reviews
Competitor pricing pages
Directories and lead lists
Job boards
Research pages
Any page where you want clean JSON instead of HTML

`extract_article`

Extract article fields from news, blog, and editorial pages.

Returns title, body text, author, and publish date when available.

`extract_metadata`

Extract page metadata including title, description, Open Graph tags, Twitter Card tags, canonical URL, and related metadata.

Why Haunt

Natural-language prompts instead of fragile CSS selectors
Browser rendering for JavaScript-heavy pages
CAPTCHA-aware extraction with machine-readable verification signals (error_code, captcha_provider, requires_human_verification)
Clean JSON output for agents, databases, and workflows
Free tier for testing