Jina AI Remote MCP Server

While some clients (like Claude Code) prefer explicit parallel_* tools (e.g., parallel_search_web), other models (like Qwen3-Next) may prefer calling singleton tools with an array of queries.

Both approaches are supported. The singleton versions of the following tools accept either a single string or an array of strings:

• search_web
• search_arxiv
• search_ssrn
• read_url

When an array is provided to a singleton tool, the server automatically executes all queries in parallel internally.

Claude Code, Claude Desktop, and Cursor enforce a fixed 25k token limit on MCP tool responses.

To prevent these clients from rejecting large responses entirely, the Jina MCP server applies a token guardrail for read_url and parallel_read_url tools:

• Single large item: Text is truncated proportionally to fit the budget.
• Multiple items: The server includes items in order until the next item would exceed the limit, then stops.

Note: Other clients like OpenAI Codex use a configurable tool_output_token_limit and do not trigger this server-side truncation.

To add the Jina MCP server to Claude Code, use the claude mcp add command.

Important: If you previously added the server using --transport sse, you must first remove it using claude mcp remove -s user jina before re-adding it with the http transport.

claude mcp add -s user --transport http jina https://mcp.jina.ai/v1 \
  --header "Authorization: Bearer ${JINA_API_KEY}"

If your MCP client supports remote servers directly, use the following configuration. The server uses Streamable HTTP transport.

Note: Some clients do not support environment variables. You may need to replace ${JINA_API_KEY} with your actual hardcoded API key (e.g., jina_xxx).

{
  "mcpServers": {
    "jina-mcp-server": {
      "url": "https://mcp.jina.ai/v1",
      "headers": {
        "Authorization": "Bearer ${JINA_API_KEY}" // optional
      }
    }
  }
}

To add the Jina MCP server to OpenAI Codex, edit your ~/.codex/config.toml file and add the following configuration:

[mcp_servers.jina-mcp-server]
command = "npx"
args = [
    "-y",
    "mcp-remote",
    "https://mcp.jina.ai/v1",
    "--header",
    "Authorization: Bearer ${JINA_API_KEY}"]

For clients that do not yet support remote MCP servers, use mcp-remote as a local proxy to connect to the remote endpoint.

{
  "mcpServers": {
    "jina-mcp-server": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.jina.ai/v1",
        "--header",
        "Authorization: Bearer ${JINA_API_KEY}"
      ]
    }
  }
}

To develop locally, clone the repository, install dependencies, and start the development server.

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

# Install dependencies
npm install

# Start development server
npm run start

To use the Jina AI Remote MCP Server, add it to your MCP client configuration (e.g., claude_desktop_config.json). The server is hosted at https://mcp.jina.ai/v1. You can optionally provide a Jina API key via the Authorization header to access advanced features.

{
  "mcpServers": {
    "jina-mcp-server": {
      "url": "https://mcp.jina.ai/v1",
      "headers": {
        "Authorization": "Bearer ${JINA_API_KEY}"
      }
    }
  }
}

{
	"mcpServers": {
	"jina-mcp-server": {
		"url": "https://mcp.jina.ai/v1",
		"headers": {
		"Authorization": "Bearer ${JINA_API_KEY}" // optional
		}
	}
	}