vavanv/my_first_mcp_server

My First MCP Server

A Model Context Protocol (MCP) server that provides tools for querying AI model information from a PostgreSQL database. This server demonstrates how to build MCP tools that interact with a database using Prisma ORM.

Features

• Database Integration: Uses Prisma ORM with PostgreSQL to store and retrieve AI company and model data
• MCP Tools: Provides tools for querying AI company information:
  • getModelName: Retrieves all AI models from the database
  • getReviews: Gets versions/reviews for a specific AI model by name
  • getChats: Retrieves all chat platforms and their information
  • getChatsByCompany: Gets chat platforms by company name
• Type Safety: Built with TypeScript for better development experience
• Database Seeding: Includes comprehensive AI company data from CSV

Database Schema

The application uses a relational schema with three main entities:

Company Model

• id: Unique identifier (CUID)
• company: Company name (unique)
• description: Company description and details
• chats: One-to-many relationship with Chat model
• llms: One-to-many relationship with LLM model
• createdAt/updatedAt: Timestamps

Chat Model

• id: Unique identifier (CUID)
• chatbot: Chatbot name
• companyId: Foreign key reference to Company
• company: Relation to Company model
• createdAt/updatedAt: Timestamps
• Unique constraint: [chatbot, companyId]

LLM Model

• id: Unique identifier (CUID)
• llm: LLM model name
• specialization: LLM specialization/description
• companyId: Foreign key reference to Company
• company: Relation to Company model
• createdAt/updatedAt: Timestamps
• Unique constraint: [llm, companyId]

Prerequisites

• Node.js (v18 or higher)
• PostgreSQL database
• Yarn or npm package manager

Installation

1. Clone the repository:

git clone <repository-url>
cd my_first_mcp_server

2. Install dependencies:

yarn install

3. Set up your environment variables by creating a .env file:

DATABASE_URL="postgresql://username:password@localhost:5432/your_database_name"

4. Generate Prisma client:

yarn prisma:generate

5. Run database migrations:

npx prisma migrate dev

6. Seed the database with AI company data from CSV:

yarn prisma:seed

Usage

Development Mode

Start the server in development mode:

yarn server:dev

Building for Production

Build the TypeScript code:

yarn server:build

Testing with MCP Inspector

Test the server using the MCP Inspector:

yarn server:inspect

For development without authentication:

yarn server:inspect2

Database Management

• Prisma Studio: Open a web interface to view and edit database data:

yarn prisma:studio

• Generate Prisma Client: After schema changes:

yarn prisma:generate

Available Tools

getModelName

Retrieves all AI models from the database.

Parameters: None

Returns: JSON array of all AI models with their properties.

getReviews

Gets versions/reviews for a specific AI model.

Parameters:

• modelName (string): The name of the model to get reviews for

Returns: JSON array of model versions for the specified model.

getChats

Retrieves all chat platforms and their information from the database.

Parameters: None

Returns: JSON array of all chat platforms with their properties.

getChatsByCompany

Gets chat platforms by company name.

Parameters:

• company (string): The company name to search for

Returns: JSON array of chat platforms for the specified company.

Sample Data

The database is populated with comprehensive AI company data including:

• 22 unique companies (OpenAI, Google, Meta AI, Anthropic, etc.)
• Chat platforms (ChatGPT, Gemini, Claude, etc.)
• LLM models with specializations (GPT-4, PaLM 2, LLaMA, etc.)

Example data structure:

OpenAI:
  - Chats: ChatGPT
  - LLMs: GPT-3, GPT-3.5, GPT-4, GPT-4o, GPT-4o mini, GPT-4.5 Orion

Google:
  - Chats: Gemini (formerly Bard)
  - LLMs: BERT, PaLM 2, Gemini, Gemma

Anthropic:
  - Chats: Claude
  - LLMs: Claude, Claude 3, Claude 4.0 Opus Thinking

Project Structure

my_first_mcp_server/
├── server/
│   ├── server.ts          # Main MCP server implementation
│   └── tools/
│       ├── index.ts       # Tool registration exports
│       ├── getCompanies.ts # Tool for getting all companies
│       ├── getChats.ts    # Tool for getting chats by company
│       └── getLLMs.ts     # Tool for getting LLMs by company
├── src/
│   └── lib/
│       └── prisma.ts      # Prisma client configuration
├── prisma/
│   ├── schema.prisma      # Database schema definition
│   ├── seed-lastone.ts    # Database seeding script for AI companies
│   ├── seed-lastone.js    # JavaScript version of seeding script
│   ├── lastone.csv        # AI company data source
│   └── migrations/        # Database migration files
├── package.json          # Project dependencies and scripts
└── tsconfig.json         # TypeScript configuration

Development

Adding New Tools

To add a new MCP tool, modify src/server.ts and add a new tool definition:

server.tool(
  "toolName",
  "Tool description",
  {
    // Parameter schema using Zod
    paramName: z.string().describe("Parameter description"),
  },
  {
    // Tool metadata
    title: "Tool Title",
    readOnlyHint: false,
    destructiveHint: false,
    idempotentHint: false,
    openWorldHint: true,
  },
  async (args) => {
    // Tool implementation
    return {
      content: [{ type: "text", text: "Result" }],
    };
  }
);

Database Changes

1. Modify prisma/schema.prisma
2. Generate a new migration:

npx prisma migrate dev --name migration_name

3. Update the Prisma client:

yarn prisma:generate

Seeding New Data

To seed the database with new AI company data:

1. Update the CSV file (prisma/lastone.csv)
2. Run the seed script:

yarn prisma:seed

Technologies Used

• MCP SDK: @modelcontextprotocol/sdk for building MCP servers
• Prisma: Database ORM for PostgreSQL
• TypeScript: Type-safe JavaScript development
• Zod: Schema validation for tool parameters
• PostgreSQL: Relational database for data storage

License

ISC

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request