mcp

Binspire MCP is an integration server that connects LLMs to the Binspire API, providing standardized tools and contextual data to power autonomous, AI-driven waste management agents. It exposes a ready-made MCP server named binspire that you run with Node.js. Once started, you can interact with the MCP server through the transport method you configured (STDIO or HTTP via a client like Claude Desktop). The server uses the Binspire API to fetch and provide contextual information and tooling to the LLM, enabling tasks such as route optimization, pickup scheduling, and waste topology queries within your agent workflows. The included configuration demonstrates how to point the MCP server to the built dist/index.js entry point and how to supply environment variables for transport and data connectivity. For Claude Desktop users, you can configure either STDIO or HTTP transport to connect to the running MCP server, passing along necessary environment variables like DATABASE_URL for data access.

Prerequisites:

• Node.js (recommended) or Bun-based setup as described in the repository
• Bun (recommended) or Node.js runtime for running the server
• Access to the Binspire API and any required credentials

Installation steps:

1. Clone the repository

git clone https://github.com/binspireai/mcp.git
cd mcp

2. Install dependencies (using Bun as per the project setup)

$ bun install

3. Copy the example environment file and configure your variables

$ cp env.example .env

4. Build if necessary (depending on your setup)

$ bun run build

5. Run the development server (STDIO)

$ bun run dev:stdio

6. Run the production server (STDIO)

$ bun run start:stdio

If you prefer HTTP transport, configure and run the HTTP variant as appropriate for your deployment environment.

Note: Ensure your environment variables (e.g., DATABASE_URL) are properly set in .env or your runtime environment before starting the server.

Tips and notes:

• The MCP server name is binspire and it expects a Node.js entry at dist/index.js. Adjust the path in the mcp configuration to match your build output.
• If you use Claude Desktop, you can connect via STDIO or HTTP transport by configuring the Claude integration with the appropriate transport and the DATABASE_URL to access your data store.
• Environment variables in the config example include TRANSPORT and DATABASE_URL. Replace with the actual values or hide sensitive data in production.
• Ensure Node.js version compatibility (the project notes specify Node.js >= 20.x). If you use Bun, ensure Bun is installed and up to date.
• For production deployments, consider containerization or hosting the node process behind a reverse proxy and secure the transport channel when using HTTP.
• If you modify the MCP entry point or build output path, update the mcpServers.binspire.args accordingly.
• Review the repository's CI workflow and security guidelines when integrating with external services.