mcpField
P
Pokémon AI Demo

Pokémon AI Demo

Local
github.com/GomesAdhikariGames & Entertainment0Updated 74w ago

About Pokémon AI Demo

Pokémon AI Demo is a full-stack app for exploring and comparing Pokémon, finding counters, and generating teams. It serves players and agent integrations through a web interface and JSON API, with team suggestions generated from natural-language descriptions.

Key features

  1. 01View Pokémon stats and abilities
  2. 02Compare two Pokémon
  3. 03Find weaknesses and counters
  4. 04Generate teams from descriptions

Use cases

  1. 01Inspect a Pokémon’s moves and evolution
  2. 02Compare type advantages between Pokémon
  3. 03Find counters to a chosen Pokémon
  4. 04Create a team for a desired play style

MCP support

The README describes Django REST API endpoints for agent integration, but does not specify MCP protocol behavior.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

Pokémon AI Demo

A full-stack application for exploring Pokémon data, comparing Pokémon, suggesting counters, and generating teams using AI. The system features a Django REST API backend and a React/MUI frontend.


Table of Contents


System Setup and Deployment

Prerequisites
  • Python 3.8+ (for backend)
  • Node.js 16+ & npm (for frontend)
  • pip (Python package manager)
Backend Setup (Django/DRF)
  1. Navigate to the backend directory:
    cd mcp_server
    
  2. Install dependencies:
    pip install -r requirements.txt
    
  3. Run migrations:
    python manage.py migrate
    
  4. Start the backend server:
    python manage.py runserver
    
    The API will be available at http://127.0.0.1:8000/api/.
Frontend Setup (React/MUI)
  1. Navigate to the frontend directory:

    cd mcp_frontend
    
  2. Install dependencies:

    npm install
    npm install @mui/material @emotion/react @emotion/styled
    npm install @mui/icons-material
    
    
    
  3. Start the frontend server:

    npm start
    

    The app will be available at http://localhost:5173/.

  4. Start the app automatically -

In terminal run command start.bat (for windows)
start.sh for (linux distro)
Environment Variables
  • By default, the frontend expects the backend at http://127.0.0.1:8000/api/agent/. Adjust API_BASE in mcp_frontend/src/App.jsx if needed.

dotenv structure (.env)

  • make a .env file inside mcpserver folder
  • contents - " POKE_API_URL = "https://pokeapi.co/api/v2/" GOOGLE_API_KEY = "your api key" "

Available Modules and Their Use

The backend exposes four main endpoints:

  1. Pokémon Info (POST /api/agent/pokemon-info/)

    • Input: { "name": "pikachu" }
    • Output: Basic info, stats, abilities, moves, evolution, and flavor text.
  2. Compare Pokémon (POST /api/agent/compare/)

    • Input: { "pokemon1": "pikachu", "pokemon2": "bulbasaur" }
    • Output: Stat-by-stat comparison, type advantage, shared/unique abilities.
  3. Suggest Counters (POST /api/agent/strategy/)

    • Input: { "name": "charizard" }
    • Output: Weaknesses and recommended counter Pokémon.
  4. Team Builder (POST /api/agent/team/)

    • Input: { "description": "balanced team with fire and water types" }
    • Output: AI-generated team with roles and images.

How to Use the Team Builder

  • Purpose: Generate a Pokémon team based on a natural language description (e.g., "offensive team with good type coverage").
  • How to Use:
    1. Go to the "Generate Team" section in the web UI.
    2. Enter your team description in the input field.
    3. Click "Generate".
    4. The AI will return a team of Pokémon, each with a name, role, and image.
  • API Usage:
    • Send a POST request to /api/agent/team/ with { "description": "your team idea" }.
    • Receive a structured team suggestion in the response.

Web Interface Walkthrough

The web app is organized into four main cards:

  1. Search for Pokémon
    • Enter a Pokémon name to view its stats, types, abilities, moves, evolution chain, and flavor text.
  2. Compare Two Pokémon
    • Enter two Pokémon names to compare their stats, type advantages, and abilities side-by-side.
  3. Suggest Counters
    • Enter a Pokémon name to see its weaknesses and recommended counters.
  4. Generate Team
    • Describe your desired team and get an AI-generated team with roles and images.

Each section provides instant feedback, loading indicators, and error messages for a smooth user experience.


Instructions for Agent Integration

You can integrate external agents or scripts with the backend API. Example usage:

  • Python Example:
    import requests
    response = requests.post('http://127.0.0.1:8000/api/agent/pokemon-info/', json={"name": "pikachu"})
    print(response.json())
    
  • Endpoints:
    • /api/agent/pokemon-info/ (POST)
    • /api/agent/compare/ (POST)
    • /api/agent/strategy/ (POST)
    • /api/agent/team/ (POST)
  • Request/Response Format: All endpoints accept and return JSON.

License

This project is for educational/demo purposes and is not affiliated with Nintendo, Game Freak, or The Pokémon Company.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.