LI
OfficialSearch

Linkup MCP Server for Natural-Language Web Search

Enable intelligent natural-language web search with the Linkup MCP server (JS) for AI assistants and developer tools.

View on GitHub
Quick Install
npx -y @LinkupPlatform/js-mcp-server

Overview

The Linkup MCP Server (JavaScript) is a lightweight Model Context Protocol (MCP) server that exposes Linkup’s web-search capabilities to MCP-compatible clients such as AI assistants, chat apps, and developer tools. It translates natural-language queries sent by an MCP client into calls to the Linkup search API, then returns structured search results and source metadata so agentic systems can reason over up-to-date web content.

Although the original JavaScript package is no longer actively maintained, it still demonstrates how to wire an MCP endpoint to Linkup’s search API and can be useful for local testing, experimentation, or as a reference. For production use, the actively maintained TypeScript/Node.js implementation is recommended (see Resources).

Features

  • Natural-language web search: accept plain-English (or other language) queries without special keyword tuning
  • Returns structured results with source citations and metadata suitable for LLM grounding
  • Works with any MCP-compatible client (websocket/HTTP-based MCP implementations)
  • Configurable API base URL and API key via CLI flags or environment variables
  • Small, easy-to-run server ideal for local development and proof-of-concepts

Installation / Configuration

Prerequisites: Node.js (12+). The JavaScript package is distributed on npm and can be run directly with npx.

Run with an API key supplied on the command line:

npx -y linkup-mcp-server --api-key=YOUR_LINKUP_API_KEY

Or set the key as an environment variable and run npx:

export LINKUP_API_KEY=YOUR_LINKUP_API_KEY
npx -y linkup-mcp-server

Clone and run locally:

git clone [email protected]:LinkupPlatform/js-mcp-server.git
cd js-mcp-server
npm install
npm start

Common CLI options

OptionDescription
–api-keyYour Linkup API key (required unless LINKUP_API_KEY env is set)
–base-urlCustom API base URL (default: https://api.linkup.so/v1)
-h, –helpShow help text

Environment variables

  • LINKUP_API_KEY — the Linkup API key to authenticate search requests
  • LINKUP_API_BASE — optional custom Linkup API base URL

Developer scripts (from repository)

ScriptPurpose
npm run buildBuild (if applicable)
npm run lintRun linter
npm run formatFormat code
npm run testRun tests

Configuration notes

  • Configure your MCP client to point to the server host and port that the MCP server exposes.
  • If you proxy or firewall the server, ensure outbound access to Linkup’s API base URL is allowed.

Available Resources

  • GitHub (JavaScript reference): https://github.com/LinkupPlatform/js-mcp-server
  • Maintained TypeScript server (recommended): https://github.com/LinkupPlatform/linkup-mcp-server
  • Linkup documentation: https://docs.linkup.so
  • Linkup API reference: https://docs.linkup.so/api-reference
  • MCP protocol specification: https://modelcontextprotocol.io

Use Cases

  1. AI assistant web grounding

    • Scenario: An LLM-based assistant needs current facts about a news event.
    • Flow: Assistant issues an MCP query like “What are the latest developments in the [topic]?” The MCP server forwards a natural-language search to Linkup, returns ranked results and URLs, and the assistant uses those sources to compose an answer with citations.

  2. Research and citation extraction

    • Scenario: A developer tool that summarises recent academic or industry research.
    • Flow: Submit a query such as “Recent research on sustainable aviation fuel” and receive summarized results plus original source links and snippets for downstream indexing.

  3. Context enrichment for conversational agents

    • Scenario: A multi-turn chat agent needs to fact-check or expand on user claims.
    • Flow: On-demand search requests allow the agent to fetch authoritative web sources and include them in the conversation context.

  4. Tooling and integrations (proof-of-concept)

    • Scenario: Integrate web search into a custom MCP-compatible client (e.g., a local desktop assistant).
    • Flow: Run the MCP server locally (via npx or from source), configure the client to connect to the local MCP endpoint, and begin sending natural-language queries.

Example natural-language queries you can try:

  • “What are current best practices for MCP server development?”
  • “How does the EU AI Act affect startups?”
  • “Summarize the latest advances in quantum computing.”

Notes and Migration

  • The javascript package in this repository is deprecated and no longer maintained. For better performance, more features, and active maintenance, use the TypeScript/Node.js implementation at https://github.com/LinkupPlatform/linkup-mcp-server.
  • This JavaScript server remains useful for learning, local testing, and as a minimal reference for integrating the MCP protocol with Linkup’s search API.

Support and Community

  • Support: [email protected]
  • Community: https://discord.com/invite/9q9mCYJa86
  • Twitter/X: https://x.com/Linkup_platform

For protocol-specific integration details, consult your MCP client’s documentation and the MCP specification to ensure compatibility with your client implementation.

Tags:search