@contentapi/mcp-server

An MCP (Model Context Protocol) server for ContentAPI — extract web content, YouTube transcripts, search the web, and more, directly from AI assistants.

Features

| Tool | Description | |------|-------------| | extract_web_content | Extract readable content from any web page (markdown or text) | | get_youtube_transcript | Get full transcript/captions from a YouTube video | | get_youtube_summary | Get an AI-generated summary of a YouTube video | | get_youtube_metadata | Get metadata (title, channel, duration, etc.) for a YouTube video | | search_web | Search the web and return results with snippets | | extract_twitter_thread | Extract a Twitter/X thread | | extract_reddit_post | Extract a Reddit post with comments | | batch_extract | Extract content from multiple URLs at once (max 10) |

Prerequisites

• Node.js ≥ 18
• A ContentAPI API key — get one at getcontentapi.com

Installation

Via npx (no install needed)

npx contentapi-mcp-server

Global install

npm install -g @contentapi/mcp-server

From source

git clone https://github.com/nicholascontentapi/contentapi-mcp-server.git
cd contentapi-mcp-server
npm install
npm run build

Configuration

The server requires a CONTENTAPI_API_KEY environment variable.

Claude Desktop

Add the following to your Claude Desktop config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Using npx

{
  "mcpServers": {
    "contentapi": {
      "command": "npx",
      "args": ["-y", "@contentapi/mcp-server"],
      "env": {
        "CONTENTAPI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Using global install

{
  "mcpServers": {
    "contentapi": {
      "command": "contentapi-mcp-server",
      "env": {
        "CONTENTAPI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Using local build

{
  "mcpServers": {
    "contentapi": {
      "command": "node",
      "args": ["/path/to/mcp-server/dist/index.js"],
      "env": {
        "CONTENTAPI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Cursor

In Cursor Settings → MCP, add a new server:

{
  "mcpServers": {
    "contentapi": {
      "command": "npx",
      "args": ["-y", "@contentapi/mcp-server"],
      "env": {
        "CONTENTAPI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Windsurf

Add to your Windsurf MCP configuration:

{
  "mcpServers": {
    "contentapi": {
      "command": "npx",
      "args": ["-y", "@contentapi/mcp-server"],
      "env": {
        "CONTENTAPI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Generic MCP Client

The server communicates over stdio using the MCP protocol. Start it as a subprocess and pass messages via stdin/stdout:

CONTENTAPI_API_KEY=your-key node dist/index.js

Or programmatically:

import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";

const transport = new StdioClientTransport({
  command: "npx",
  args: ["-y", "@contentapi/mcp-server"],
  env: {
    CONTENTAPI_API_KEY: "your-api-key",
  },
});

const client = new Client({ name: "my-app", version: "1.0.0" });
await client.connect(transport);

// List tools
const tools = await client.listTools();

// Call a tool
const result = await client.callTool({
  name: "extract_web_content",
  arguments: { url: "https://example.com" },
});

Tool Reference

extract_web_content

Extract readable content from any web page.

| Parameter | Type | Required | Default | Description | |-----------|------|----------|---------|-------------| | url | string | ✅ | — | Web page URL | | format | "markdown" | "text" | ❌ | "markdown" | Output format | | ignore_robots | boolean | ❌ | false | Ignore robots.txt |

get_youtube_transcript

Get transcript/captions from a YouTube video.

| Parameter | Type | Required | Default | Description | |-----------|------|----------|---------|-------------| | url | string | ✅ | — | YouTube video URL | | lang | string | ❌ | auto | Language code (e.g. "en", "pt") |

get_youtube_summary

Get an AI-generated summary of a YouTube video.

| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | url | string | ✅ | YouTube video URL |

get_youtube_metadata

Get video metadata (title, description, channel, duration, etc.).

| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | url | string | ✅ | YouTube video URL |

search_web

Search the web.

| Parameter | Type | Required | Default | Description | |-----------|------|----------|---------|-------------| | query | string | ✅ | — | Search query | | count | number (1-10) | ❌ | 5 | Number of results |

extract_twitter_thread

Extract a Twitter/X thread.

| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | url | string | ✅ | Twitter/X thread URL |

extract_reddit_post

Extract a Reddit post with comments.

| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | url | string | ✅ | Reddit post URL |

batch_extract

Extract content from multiple URLs at once.

| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | urls | string[] (max 10) | ✅ | Array of URLs |

Error Handling

The server returns clear error messages for common issues:

• Missing API key — Prompts you to set CONTENTAPI_API_KEY
• Invalid URL — Validation errors from Zod schemas
• API errors — HTTP status code + error message from ContentAPI
• Network errors — Connection/timeout errors surfaced to the client

Development

# Install dependencies
npm install

# Build
npm run build

# Watch mode
npm run dev

# Run locally
CONTENTAPI_API_KEY=your-key npm start

License

MIT