Kagi

Provides web search via the Kagi API with a dedicated MCP tool for parallel queries.

typescript

2

GitHub Stars

typescript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "elliottlawson-kagi-mcp-server": {
      "command": "npx",
      "args": [
        "github:elliottlawson/kagi-mcp-server"
      ],
      "env": {
        "KAGI_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

This MCP server provides web search capabilities through the Kagi API and exposes a dedicated tool for secure, parallel queries. You can run it locally or via a client, then invoke the kagi_web_search tool to retrieve and format search results for use in your AI workflows.

How to use

You run the server as a local component that your MCP client talks to via stdio. Once running, call the kagi_web_search tool from your MCP-enabled assistant to perform web searches and receive structured results. The tool is designed to handle multiple queries in parallel and returns formatted results suitable for downstream reasoning.

How to install

Prerequisites you need to meet before running the server:

Node.js 18 or higher
Kagi API key

Quick Start (no installation) you can run directly with npx:

KAGI_API_KEY=your_api_key_here npx github:elliottlawson/kagi-mcp-server

Installation steps if you want to run from a cloned copy:

git clone https://github.com/elliottlawson/kagi-mcp-server.git
cd kagi-mcp-server
KAGI_API_KEY=your_api_key_here node build/index.js

If you need to modify code and rebuild, install dependencies and build the project:

npm install
npm run build

Starting the server after a build (examples assume completed build):

npm start

Additional configuration and usage notes

Configure clients with explicit MCP server entries to ensure the client can communicate with the server over stdio and share the required environment variable.

Debugging and troubleshooting

If you want to inspect MCP activity and verify the server is receiving requests, use the MCP Inspector tooling to debug the server while it runs.

npx @modelcontextprotocol/inspector npx github:elliottlawson/kagi-mcp-server

Or, if you have a cloned copy, run:

npx @modelcontextprotocol/inspector node build/index.js

Then open MCP Inspector at http://localhost:5173 and provide your KAGI_API_KEY in the inspector settings.


## Available tools

### kagi\_web\_search

A dedicated MCP tool that performs web searches through the Kagi API, supports multiple queries in parallel, and returns formatted results for use by an AI assistant.