home / mcp / jina ai remote mcp server

Jina AI Remote MCP Server

Provides access to Jina Reader, Embeddings and Reranker APIs along with web, image, and markdown tools via MCP

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "acchuang-jina-mcp": {
      "url": "https://mcp.jina.ai/sse",
      "headers": {
        "JINA_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

You can access Jina’s MCP-enabled remote services through a dedicated MCP server that exposes Reader, Embeddings, and Reranker APIs, plus a set of URL-to-markdown, search, and deduplication tools. This server lets you connect with a simple, standardized interface from your MCP-enabled clients to perform web content extraction, search, image handling, and semantic deduplication at scale.

How to use

You connect your MCP-enabled client to the remote server using either the HTTP endpoint or a local proxy that talks to the remote server. Once connected, you can perform tasks such as extracting clean content from web pages, searching the web and arXiv, finding images, and reordering or deduplicating results with embeddings-based tools. If your client supports direct MCP connections, use the remote HTTP URL with an optional API key for higher rate limits. If your client only supports a local proxy, run the local proxy to reach the remote MCP endpoint.

Key capabilities include content extraction from pages (read_url and capture_screenshot_url), broad web and arXiv search (search_web, search_arxiv, search_image), and result refinement or deduplication (sort_by_relevance, deduplicate_strings, deduplicate_images). For best performance, supply your Jina API key where required and enable the optional tools according to your usage pattern.

Two connection methods are provided. The HTTP method connects directly to the remote endpoint. The stdio method uses a local proxy to forward requests to the remote server. Choose the method your client supports and configure the appropriate headers or proxy command.

How to install

Prerequisites: ensure you have Node.js and npm installed on your system.

Install dependencies and start the development server if you are developing or testing locally.

# Clone the repository
git clone https://github.com/jina-ai/MCP.git
cd MCP

# Install dependencies
npm install

# Start development server
npm run start

Configuration and connection guides

You can connect using either the HTTP endpoint or the local stdio proxy. Below are the exact configuration snippets you can reuse in your MCP client setup.

{
  "mcpServers": {
    "jina_mcp_server": {
      "url": "https://mcp.jina.ai/sse",
      "headers": {
        "Authorization": "Bearer ${JINA_API_KEY}"
      }
    }
  }
}

Alternative local proxy configuration

If your client does not support remote MCP servers yet, you can connect through a local proxy that forwards requests to the remote MCP server.

{
  "mcpServers": {
    "jina_mcp_server": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.jina.ai/sse",
        "--header",
        "Authorization: Bearer ${JINA_API_KEY}"
      ]
    }
  }
}

Available tools

read_url

Extract clean, structured content from web pages as markdown using the Reader API (optional API key)

capture_screenshot_url

Capture high-quality screenshots of web pages via the Reader API (optional API key)

search_web

Search the web for current information and news via the Reader API (requires API key)

search_arxiv

Search academic papers on arXiv via the Reader API (requires API key)

search_image

Search for web images via the Reader API (requires API key)

sort_by_relevance

Rerank documents by relevance to a query via the Reranker API (requires API key)

deduplicate_strings

Get top-k semantically unique strings via Embeddings API with submodular optimization (requires API key)

deduplicate_images

Get top-k semantically unique images via Embeddings API with submodular optimization (requires API key)