🤖 Multi-Agent + Multi-Tool Backend Builder System

A sophisticated system that combines multiple AI agents with multiple tools (MCP + Custom Tools) to automatically design, generate, and manage production-ready backend applications. Built with AutoGen, Gemini, and Groq APIs, this system showcases advanced capabilities in both multi-agent collaboration and multi-tool utilization.

Note: This project extends beyond a standard multi-tool agent implementation by integrating multiple specialized agents that can leverage multiple tools in a coordinated fashion, creating a more powerful and flexible system than either approach alone.

🎯 Overview

This system employs four specialized AI agents that collaborate to build complete backend applications:

• 🧠 Architect Agent: Plans system architecture and design
• 💻 Coder Agent: Implements code based on architecture plans
• ⚙️ Ops Agent: Handles deployment, testing, and operations
• 🔍 Reviewer Agent: Ensures production readiness and quality

✨ Key Features

Multi-Agent System

• Specialized Agents: Four distinct AI agents with specific roles and responsibilities
• Collaborative Workflow: Agents work together in a coordinated pipeline
• Role-Based Access: Each agent has specific permissions and tool access
• Distributed Knowledge: Different agents maintain different aspects of project state

Multi-Tool Capabilities

• MCP Server Integration: Seamless interaction with external MCP servers
• Custom Tools: Specialized tools for code generation, validation, and testing
• Tool Chaining: Ability to combine multiple tools in sequence
• Dynamic Tool Selection: Agents can choose the right tool for each task

Development Features

• Production-Ready Code: Enforces comprehensive production checklist
• Incremental Development: 20% phases with user feedback loops
• Multi-LLM Strategy: Gemini for reasoning, Groq for fast code generation
• Security-First: Built-in security validation and best practices
• Quality Assurance: Automated code review and testing
• Modular Architecture: Easily extensible agent system
• Real-time Collaboration: Agents work together to solve complex tasks
• Comprehensive Testing: Built-in test generation and execution

🛠️ System Capabilities

What It Can Create

• RESTful APIs: Complete backend services with proper endpoints
• Database Schemas: SQL and NoSQL database designs
• Authentication Systems: Secure user management and access control
• WebSocket Services: Real-time communication endpoints
• Background Tasks: Asynchronous job processing
• API Documentation: Interactive API documentation (Swagger/OpenAPI)
• Test Suites: Comprehensive test coverage
• Docker Configurations: Containerization setup
• CI/CD Pipelines: Automated testing and deployment workflows

Tools Integration

• MCP Filesystem: For file operations within the workspace
• MCP Terminal: For executing shell commands
• Code Analysis Tools: For quality and security checks
• Version Control: Git integration for code management
• Dependency Management: Automatic requirements tracking

🏗️ Architecture

Multi-Agent System/
├── agents/                 # AI Agent implementations
│   ├── architect.py       # System design & planning
│   ├── coder.py          # Code implementation
│   ├── ops.py            # Operations & testing
│   └── reviewer.py       # Quality assurance
├── config/               # Configuration management
│   ├── gemini_config.json
│   ├── groq_config.json
│   └── tools_config.json
├── utils/                # Utilities
│   └── mcp_client.py     # MCP server integration
├── workspace/            # Generated projects go here
├── main.py              # Main orchestrator
└── production_checklist.md # Quality standards

🚀 Quick Start

1. Prerequisites

• Python 3.8+
• Virtual environment (recommended)
• Gemini API key
• Groq API key (optional but recommended)

2. Installation

# Clone or navigate to the project directory
cd multi-tool-agent

# Install dependencies
pip install -r requirements.txt

# Install MCP Terminal server
npm install -g @dillip285/mcp-terminal

3. Configuration

1. Set up API keys in .env:

# Copy and edit the .env file
cp .env .env.local

# Edit .env.local with your actual API keys
GEMINI_API_KEY=your_gemini_api_key_here
GROQ_API_KEY=your_groq_api_key_here

2. MCP Server Setup: The system uses external MCP servers:
   • Filesystem MCP: https://www.claudemcp.com/servers/filesystem
   • Terminal MCP: @dillip285/mcp-terminal (locally hosted)

4. Usage

Basic Usage

python main.py --project "AI Chat Backend" --description "Build a FastAPI backend for AI-powered chat with user authentication"

Full Build (No Incremental Feedback)

python main.py --project "Content Generator API" --description "AI content generation service with rate limiting" --full

📋 Production Checklist

The system enforces a comprehensive production checklist covering:

• ✅ Core Application: Environment variables, dependencies, project structure
• ✅ Code Quality: PEP8, error handling, logging, modularization
• ✅ Security: Input validation, authentication, secrets management
• ✅ LLM Safeguards: Prompt templates, output validation, cost tracking
• ✅ Testing: Unit tests, integration tests, coverage goals
• ✅ Observability: Structured logging, metrics, health checks
• ✅ Performance: Async patterns, caching, efficient queries
• ✅ Deployment: Docker configs, environment management
• ✅ Documentation: README, API docs, code documentation

🔄 Workflow

Incremental Development (Default)

1. Architecture Planning (20%)

   • Architect Agent designs system structure
   • User reviews and approves plan

2. Core Implementation (40%)

   • Coder Agent implements basic structure
   • Ops Agent sets up environment
   • User tests basic functionality

3. API & Features (60%)

   • Complete API implementation
   • Add advanced features
   • User validates functionality

4. Testing & Quality (80%)

   • Comprehensive testing setup
   • Code quality validation
   • Security review

5. Production Readiness (100%)

   • Final production checklist validation
   • Deployment preparation
   • Documentation completion

Agent Responsibilities

🧠 Architect Agent

• Role: Senior backend architect
• Tools: Gemini API for complex reasoning
• Outputs: System design, API specifications, database schema

💻 Coder Agent

• Role: Backend developer
• Tools: Groq (fast) + Gemini (complex), Filesystem MCP
• Outputs: FastAPI applications, database models, configuration files

⚙️ Ops Agent

• Role: DevOps engineer
• Tools: Terminal MCP
• Outputs: Test execution, server management, deployment scripts

🔍 Reviewer Agent

• Role: Senior code reviewer
• Tools: Gemini API, Filesystem MCP
• Outputs: Code quality reports, security analysis, production readiness validation

🛠️ Configuration

LLM Configuration

Gemini (Complex reasoning tasks):

• Architecture planning
• Code review and analysis
• Security validation
• Complex problem solving

Groq (Fast code generation):

• Code implementation
• Simple transformations
• Quick responses

MCP Server Configuration

The system integrates with external MCP servers for secure operations:

{
  "mcp_servers": {
    "filesystem": {
      "url": "https://www.claudemcp.com/servers/filesystem",
      "workspace_path": "./workspace"
    },
    "terminal": {
      "command": "npx",
      "args": ["@dillip285/mcp-terminal", "--allowed-paths", "./workspace"]
    }
  }
}

📁 Generated Projects

All generated projects are created in the workspace/ directory with:

• FastAPI application with proper structure
• Database models (SQLAlchemy)
• API endpoints with validation
• Configuration management
• Testing setup
• Docker configuration
• Documentation

Example generated structure:

workspace/ai-chat-backend/
├── main.py              # FastAPI application
├── app/
│   ├── __init__.py
│   ├── config.py        # Configuration management
│   ├── models.py        # Database models
│   └── routers/         # API endpoints
├── tests/               # Test suite
├── requirements.txt     # Dependencies
├── .env.example        # Environment template
├── Dockerfile          # Container configuration
└── README.md           # Project documentation

🔒 Security Features

• Input Validation: All endpoints include proper validation
• Authentication: JWT-based auth when needed
• Rate Limiting: Built-in rate limiting for APIs
• Security Headers: Proper CORS and security middleware
• Secrets Management: Environment-based configuration
• SQL Injection Prevention: Parameterized queries

🧪 Testing

The system generates comprehensive test suites:

• Unit Tests: Core logic testing
• Integration Tests: API endpoint testing
• Security Tests: Vulnerability scanning
• Performance Tests: Load testing setup

Run tests for generated projects:

cd workspace/your-project
pytest -v

📊 Monitoring & Observability

Generated applications include:

• Health Check Endpoints: /health, /ready
• Structured Logging: JSON formatted logs
• Metrics Collection: Request/response metrics
• Error Tracking: Comprehensive error handling

🚀 Deployment

Generated projects are deployment-ready with:

• Docker Support: Multi-stage Dockerfiles
• Environment Configuration: Dev/staging/prod configs
• Database Migrations: Alembic setup
• CI/CD Ready: GitHub Actions templates

🔧 Troubleshooting

Common Issues

1. MCP Server Connection Failed

   # Ensure MCP terminal server is installed
   npm install -g @dillip285/mcp-terminal
   
   # Check if server is accessible
   npx @dillip285/mcp-terminal --help

2. API Key Issues

   # Verify API keys are set
   echo $GEMINI_API_KEY
   echo $GROQ_API_KEY

3. Permission Errors

   # Ensure workspace directory is writable
   chmod 755 workspace/

Debug Mode

Enable detailed logging:

export LOG_LEVEL=DEBUG
python main.py --project "Test Project" --description "Test description"

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests for new functionality
5. Submit a pull request

📄 License

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

🙏 Acknowledgments

• AutoGen Framework: Multi-agent orchestration
• Google Gemini: Advanced reasoning capabilities
• Groq: Fast inference for code generation
• MCP Protocol: Secure tool integration
• FastAPI: Modern Python web framework

---

Built with ❤️ by the Multi-Agent Backend Builder System

For support or questions, please open an issue in the repository.