ai-agent

You can run an MCP server in NodeJS that exposes REST resources for users and messages and provides an MCP client to access those resources. This setup enables you to query, read, and interact with external resources through a standardized protocol, making it easier to integrate with other MCP-enabled tools and workflows.

How to use

You can access resources in two practical ways: use the MCP client to connect and list or read resources, or access REST endpoints directly from your preferred HTTP client. The server exposes a users resource at api://users and a simple hello resource at hello://world for quick interaction. Use the client to discover what resources are available and read the content you need.

How to install

Prerequisites you need to have before starting:

• Node.js installed (compatible with AbortController features)
• npm installed and configured
• TypeScript tooling as needed
• A terminal or shell access

Choose one of the following installation paths to run the MCP server locally.

Option A — Clone the project, install dependencies, and run the server

git clone https://github.com/moises-paschoalick/ai-agent-with-mcp.git
cd ai-agent-with-mcp
npm run setup
npm start

Option B — Install dependencies and run directly (development mode)

npm install
npm run build
npm start

Option C — Run in development mode with dev script

If you prefer a development workflow, you can start the server in development mode using the dev script.

npm run dev

Cursor integration note

Cursor users can configure an MCP client to access the server resources. You can install the Cursor client, then configure an MCP server entry to run the client’s access tool and connect to your MCP server.

Available tools

listResources

Lists all resources available on the MCP server so you can discover what endpoints you can query or read.

readResource

Reads the content of a specific MCP resource by URI, such as api://users to get user data.