home / mcp / algtools mcp server

Algtools MCP Server

Remote MCP server hosting AI-powered tools to search cursor rules and AlgtoolsUI components via Cloudflare.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "algtools-mcp": {
      "url": "https://remote-mcp-server-authless.<your-account>.workers.dev/sse",
      "headers": {
        "AI_SEARCH_API_TOKEN": "your-token-here"
      }
    }
  }
}

You can deploy a remote MCP server that hosts AI-powered tools for searching cursor rules and AlgtoolsUI component information. It runs on Cloudflare Workers and exposes an MCP JSON-RPC endpoint you can connect to from MCP clients, playgrounds, or local proxies. This guide shows how to use the server, install it locally or remotely, and configure access to its endpoints and environment variables.

How to use

Connect to the remote MCP server from your MCP client or testing playground. You will access the MCP HTTPS endpoint at the server’s /mcp path and you can test the available tools through the provided testing interface. Use the public web interface to explore the API, try out endpoints, and verify tool behavior before integrating into your own applications.

How to install

Prerequisites you need before starting:

Node.js and npm or pnpm installed on your system
Access to Cloudflare Workers setup (Wrangler) if you plan to deploy locally or configure environment secrets

To run a remote MCP server instance that doesn’t require authentication, follow these steps.

1) Deploy the remote MCP server to a Cloudflare Workers URL (example shown is a Deploy URL you can expect after deployment):

https://remote-mcp-server-authless.<your-account>.workers.dev/sse

Additional sections

Environment variables and secrets are used by this MCP server. Access them inside your tools via the server environment object. For example, you can read an API token from environment variables to enable AI-powered search features.

async init() {
  this.server.tool("myTool", {}, async () => {
    // Access environment variables through this.env
    const apiToken = this.env.AI_SEARCH_API_TOKEN;
    // ...
  });
}

Setting up secrets is recommended. Use Wrangler secrets for sensitive values rather than plain environment variables. Example:

# Set a secret
wrangler secret put AI_SEARCH_API_TOKEN

# For local development, create a .dev.vars file:
echo "AI_SEARCH_API_TOKEN=your-token-here" > .dev.vars

You can test the MCP server with the built-in web interface. For local testing, open the Scalar UI at http://localhost:8787/docs and review the OpenAPI spec at http://localhost:8787/openapi.json. This interface supports interactive API documentation, request testing, and schema exploration.

Available tools

cursorRules

Performs AI-powered search on the Algenium team's cursor rules using Cloudflare AutoRAG. Takes a query string and returns relevant cursor rules for quick reference.

algtoolsUI

Performs AI-powered search on AlgtoolsUI component information using Cloudflare AutoRAG. Takes a query string and returns relevant component data for rapid prototyping.