AI Assistant

Overview

AI Assistant is a modern full-stack application with multi-modal capabilities, supporting multiple language models (LLMs) and tool integrations. It provides an intuitive interface for interacting with AI models and supports various specialized tools such as math calculations, weather queries, test scenario generation, and more.

Key Features

• Multi-LLM Support: Integration with OpenAI, Google, Anthropic, and NVIDIA models
• MCP Tool Integration: Specialized tools via the Model Control Protocol (MCP)
• Responsive Design: Modern UI for both desktop and mobile devices
• Authentication System: Secure user authentication and session management
• Multilingual Support: English and Chinese interfaces
• Real-time Chat: Fluid conversation experience
• File Upload: Document and image upload support for AI analysis and processing
• Test Scenario Generation: Automatically generate test scenarios for API endpoints

System Architecture

AI Assistant uses a three-tier architecture:

1. Frontend: React application built with Next.js

   • Responsive design with Tailwind CSS
   • Internationalization support
   • Real-time chat interface
   • File upload and management components

2. Backend: Python service using FastAPI

   • RESTful API
   • Multi-LLM integration
   • MCP tool management
   • User authentication
   • File upload handling

3. Database: PostgreSQL

   • User management
   • Session storage
   • Tool configurations
   • File metadata storage

Installation

Quick start:

# Clone the repository
git clone https://github.com/yourusername/ia-assistant.git
cd ia-assistant

# Setup backend
cd backend
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

# Setup frontend
cd ../frontend
npm install

For detailed installation instructions, refer to the Installation Guide.

MCP Tool Integration

AI Assistant integrates various specialized tools through the Model Control Protocol (MCP):

Available Tools

• Math Tool: Perform mathematical calculations
• Weather Tool: Retrieve weather information
• Test Scenario Generator: Generate comprehensive test scenarios for APIs
• DuckDuckGo Search: Search the web for information
• More Tools: Continuously being integrated...

MCP Architecture

The MCP tools are located in the mcp_server directory, with each tool in its own subdirectory:

• mcp_server/math/: Math calculation tool
• mcp_server/weather/: Weather information tool
• mcp_server/TestScenarioGenerator/: Test scenario generation tool
• mcp_server/duckduckgo/: Web search tool

Each tool follows the MCP protocol and can be used individually or through the integrated backend.

Start MCP Server

cd backend
python mcp_server.py

Test Scenario Generator

The Test Scenario Generator is a specialized tool for automatically creating test scenarios for API endpoints:

Features

• Generate test scenarios from API descriptions
• Support for various HTTP methods and request/response types
• Export test cases in different formats
• Integration with testing frameworks

Usage Example

{
  "name": "generate_test_scenarios_from_description",
  "input": {
    "api_description": "User registration API that accepts username, password, email, and returns success or error response",
    "api_path": "/api/v1/users/register"
  }
}

File Upload Feature

AI Assistant supports uploading various file formats:

Supported File Types

• Documents: PDF, DOCX, TXT, MD, CSV, JSON
• Images: JPG, JPEG, PNG, GIF, BMP, WEBP

How to Use

1. Click the upload button in the chat interface
2. Select the file type (document or image)
3. Choose a file from your local storage
4. Once uploaded, the file will appear in the chat, and the AI can analyze its content

Usage Guide

1. Access the frontend: http://localhost:3000
2. Create an account or log in
3. Start chatting, for example:
   • "Calculate 123 + 456"
   • "What's the weather in New York?"
   • "Generate test scenarios for a login API"
   • "Search for information about climate change"
   • "Upload a file and analyze its content"

Development

Directory Structure

ia-assistant/
├── backend/           # FastAPI backend service
│   ├── app/          # Application code
│   ├── tests/        # Test files
│   └── requirements.txt
├── frontend/         # Next.js frontend application
│   ├── src/         # Source code
│   ├── public/      # Static assets
│   └── package.json
├── mcp_server/       # MCP tool servers
│   ├── math/        # Math tool
│   ├── weather/     # Weather tool
│   ├── TestScenarioGenerator/  # Test scenario generator
│   └── duckduckgo/  # Web search tool
└── README.md        # Project documentation

Configuration

Backend Configuration

In backend/.env:

DATABASE_URL=postgresql://user:password@localhost:5432/ia_assistant
OPENAI_API_KEY=your_openai_key
GOOGLE_API_KEY=your_google_key
ANTHROPIC_API_KEY=your_anthropic_key
DEFAULT_LLM_PROVIDER=google
DEFAULT_GOOGLE_MODEL=gemini-2.0-flash

Tool Configuration

Each tool may have its own .env file for configuration. For example, in mcp_server/TestScenarioGenerator/.env:

TEST_GENERATOR_DEFAULT_LLM_PROVIDER=google
TEST_GENERATOR_GOOGLE_API_KEY=your_google_api_key
TEST_GENERATOR_DEFAULT_GOOGLE_MODEL=gemini-2.0-flash

Frontend Configuration

In frontend/.env.local:

NEXT_PUBLIC_API_URL=http://localhost:8000

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgements

Thanks to the following technologies and services:

• Next.js and React
• FastAPI
• PostgreSQL
• AI services from OpenAI, Google, Anthropic, and NVIDIA
• Tailwind CSS
• LangChain and LangGraph for LLM orchestration
• Model Control Protocol (MCP) for tool integration