home / mcp / jina ai mcp server

Jina AI MCP Server

Provides access to Jina AI web services via Claude including read, search, and fact-check tools.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "joebuildsstuff-mcp-jina-ai": {
      "command": "npx",
      "args": [
        "-y",
        "jina-ai-mcp-server"
      ],
      "env": {
        "JINA_API_KEY": "<YOUR_KEY>"
      }
    }
  }
}

You can run a Jina AI MCP Server to access Jina AI web services through Claude, enabling you to read and extract web content, perform web searches, and fact-check statements. This MCP server ties these tools into your Claude workflow with simple local or NPX-based startup options.

How to use

You connect with an MCP client from your Claude Desktop setup by adding an MCP server entry. Once configured, you can invoke three main tools: read_webpage to extract content from pages, search_web to query Jina AI’s search API, and fact_check to verify statements against grounding data. Each tool returns structured results that include content formatted for LLMs, along with metadata such as titles, descriptions, and references. Use the tools to enrich your conversations, verify facts, and pull in relevant web content on demand.

How to install

Prerequisites you need before starting are simple: you must have Node.js installed and an API key from Jina AI. Prepare your Claude Desktop config with one or more MCP server entries using either the NPX-based startup or a local installation path.

{
  "mcpServers": {
    "jina-ai-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "jina-ai-mcp-server"
      ],
      "env": {
        "JINA_API_KEY": "<YOUR_KEY>"
      }
    }
  }
}

If you prefer a local installation, you can run the server from a built distribution. The configuration snippet below shows starting with Node and pointing to the built index file. Replace the path with the actual dist/index.js after building.

{
  "mcpServers": {
    "jina-ai-mcp-server": {
      "command": "node",
      "args": [
        "/path/to/jina-ai-mcp-server/dist/index.js"
      ],
      "env": {
        "JINA_API_KEY": "<YOUR_KEY>"
      }
    }
  }
}

Additional setup notes

Configuration files for macOS and Windows show where Claude Desktop stores the MCP configuration. You can place the config in the standard Claude Desktop config path used by your platform. If you encounter debugging issues, MCP servers communicate over stdio, and you can use an inspector tool to access debugging information in your browser.

Security and troubleshooting

Keep your JINA_API_KEY secure. Do not expose it in shared configs. If you run into startup issues, verify that your environment variable is set correctly in the config you load into Claude Desktop, and ensure the server process has access to the dist/index.js path when using the local installation method.

Available tools

read_webpage

Extracts content from web pages and formats it for LLMs. Supports multiple output formats and options to include links/images, and generate image alt text with cache control.

search_web

Performs web searches using Jina AI APIs. Returns a structured list with titles, descriptions, and content. Supports configurable number of results and multiple output formats.

fact_check

Fact-checks statements using Jina AI grounding. Provides factuality scores, supporting evidence, and optional deep-dive analysis with references and quotes.