Multi-agent-orchestrator-system-with-standardized-A2A-Protocol

<p align="center"> <img src="./banner.png" alt="Multi-Agent Orchestrator System Banner" width="100%" /> </p> # Multi-Agent Orchestrator System A production-ready multi-agent system with an orchestrator pattern for coordinating specialized agents. Features intelligent routing, context-aware conversations, and a beautiful interactive dashboard with real-time communication flow visualization. ## ✨ Features ### Core Capabilities - **Orchestrator Agent**: Coordinates queries across all agents and aggregates responses - **Agent-to-Agent Protocol**: Standardized communication framework (A2A Protocol v1.0) - **Independent Agents**: Each agent has separate vector database storage - **Smart Routing**: Intelligently routes queries only to relevant agents - **Context Awareness**: Remembers conversation history to infer user context - **General Question Handling**: Answers non-schedule questions directly without agent coordination ### AI Integration - **Multi-API Support**: Uses DeepSeek and Gemini APIs for parallel processing - **Intelligent Query Detection**: Automatically detects schedule vs general questions - **Dynamic Suggestions**: LLM-generated follow-up suggestions based on context - **Concise Responses**: Optimized prompts for brief, clear answers ### Web Dashboard - **Interactive Chat Interface**: Real-time chat with typing indicators - **Animated Communication Flow**: Visual representation of agent communication with SVG paths - **Real-time Status Monitoring**: Live agent status and query tracking - **Communication Log**: Detailed log of all agent-to-agent communications - **Modern UI**: Black and light blue theme inspired by InfinityFX - **Responsive Design**: Works seamlessly on desktop and mobile ### Deployment - **Railway Ready**: Pre-configured for Railway deployment - **Environment Detection**: Automatically adapts to deployment environment - **Production Optimized**: Error handling, retries, and rate limit management ### Communication Flow The system features a beautiful animated communication flow diagram that shows: - **Request Flow**: Blue paths showing queries from user to orchestrator to agents - **Response Flow**: Green dashed paths showing responses back to user - **Real-time Animation**: Nodes pulse and paths animate during processing - **Status Indicators**: Visual feedback for each stage of communication ## 🚀 Quick Start ### Prerequisites - Python 3.9+ - Google Gemini API key (required) - DeepSeek API key (optional, for faster parallel processing) ### Installation 1. **Clone the repository:** ```bash git clone https://github.com/Anujpatel04/Multi-agent-orchestrator-system-with-standardized-A2A-Protocol.git cd Multi-agent-orchestrator-system-with-standardized-A2A-Protocol ``` 2. **Create virtual environment:** ```bash python -m venv venv source venv/bin/activate ``` 3. **Install dependencies:** ```bash pip install -r requirements.txt ``` 4. **Configure environment:** Create a `.env` file with your API keys: ```bash GEMINI_API_KEY=your_gemini_api_key_here GEMINI_API_KEY_AGENT1=your_gemini_api_key_here DEEPSEEK_API_KEY_ORCHESTRATOR=your_deepseek_key DEEPSEEK_API_KEY=your_deepseek_api_key_here ``` **Note:** - Orchestrator uses DeepSeek API for routing and aggregation - Agent 1 uses Gemini API - Agent 2 uses DeepSeek API (different from Agent 1) for better parallel processing - If DeepSeek keys are not provided, components will fallback to Gemini - Using different APIs allows all components to process queries in parallel, making the system much faster! 5. **Run the web application:** ```bash python web_app.py ``` 6. **Access the dashboard:** Open your browser and navigate to: `http://localhost:5001` ## 💡 Usage ### Web Interface Start the web server: ```bash python web_app.py ``` The dashboard provides: - **Chat Interface**: Natural language queries with intelligent routing - **Real-time Agent Status**: Monitor agent availability and schedules - **Animated Flow Diagram**: Visualize communication flow in real-time - **Communication Log**: Track all agent interactions - **Dynamic Suggestions**: AI-generated follow-up questions ### Query Types 1. **All Agents**: Query all agents simultaneously 2. **Smart Routing**: Intelligently route to relevant agents only 3. **Find Common Time**: Agent-to-agent communication to find common free time ### Context Awareness The orchestrator remembers conversation context: - **Example 1**: - User: "What is user 1's schedule?" - User: "When are they free?" (orchestrator infers "user 1") - **Example 2**: - User: "What is the weather today?" (answered directly, no agent coordination) - User: "What is user 1's schedule?" (routes to Agent 1) ### CLI Interface ```bash python orchestrator_app.py ``` Options: 1. Query all agents 2. Smart routing 3. Find common free time 4. View system summary ### Scripts Populate sample schedules: ```bash python scripts/populate_routines.py ``` Query agents separately: ```bash python scripts/query_agents.py ``` ## Agent-to-Agent Protocol The system implements a standardized A2A Protocol v1.0 for inter-agent communication: - Standardized message format - Message types (Query, Response, Broadcast, etc.) - Agent registry for discovery - Message routing and history - Priority levels and correlation IDs See `docs/PROTOCOL.md` for full specification. ## UI Features ### Communication Flow Diagram - **Interactive Nodes**: Hover effects and click interactions - **Animated Paths**: SVG paths that pulse and flow during communication - **Status Indicators**: Real-time status updates (Idle, Processing, Active) - **Visual States**: Active (blue), Processing (orange), Success (green) - **Query Counter**: Badge showing number of processed queries ### Environment Variables Set these in your Railway project: - `GEMINI_API_KEY` - `GEMINI_API_KEY_AGENT1` - `DEEPSEEK_API_KEY_ORCHESTRATOR` - `DEEPSEEK_API_KEY` - `PORT` ## 🛠️ Technology Stack - **Backend**: Python 3.9+, Flask - **AI**: Google Gemini 2.5 Pro, DeepSeek Chat - **Database**: ChromaDB (vector database) - **Frontend**: HTML5, CSS3, JavaScript (ES6+) - **Communication**: REST API, A2A Protocol - **Deployment**: Railway, Heroku-compatible ## 📝 Recent Updates ### v2.0 - Enhanced Intelligence & UI - ✅ Context-aware user inference from conversation history - ✅ General question handling (direct LLM responses) - ✅ Enhanced communication flow diagram with animations - ✅ DeepSeek API integration for parallel processing - ✅ Improved response formatting (no markdown) - ✅ Dynamic LLM-generated follow-up suggestions - ✅ Real-time status indicators and animations - ✅ Railway deployment support ### v1.0 - Initial Release - ✅ Multi-agent orchestrator system - ✅ A2A Protocol implementation - ✅ Web dashboard - ✅ Schedule management - ✅ Agent-to-agent communication ## 🤝 Contributing Contributions are welcome! Please feel free to submit a Pull Request. - Google Gemini API for LLM capabilities - DeepSeek for additional LLM support - ChromaDB for vector database functionality - InfinityFX for UI design inspiration