AgentWeaver

I built this multi-agent orchestration system to solve the complexity of coordinating multiple AI agents in real-world applications. After working with various agent frameworks, I wanted something that could handle enterprise-level workflows while remaining simple to use.

What I Created

This system represents my solution to several key challenges I encountered:

• Agent Coordination: Managing multiple AI agents that work together on complex tasks
• Real-time Dashboard: Live WebSocket-powered interface to see exactly what your agents are doing
• State Management: Persistent storage with Redis that survives restarts and crashes
• Workflow Orchestration: Linear workflows with text analysis, data processing, and API integration
• Production Ready: Built to handle real workloads with proper error handling and monitoring

Getting Started

I've made setup as straightforward as possible:

# Install dependencies
pip install -r requirements.txt

# Launch the backend server
python main.py

# In another terminal, start the React dashboard
cd frontend
npm install
npm run dev

Your system will be running at:

• Backend API: http://localhost:8000
• Live Dashboard: http://localhost:3000

See It In Action

I've included working demos that showcase the real-time features:

# Watch your dashboard come alive with real agent activity
python examples/demos/dashboard_demo.py

Open http://localhost:3000 and you'll see:

• Agents changing status in real-time (idle → busy → idle)
• Workflow progress bars moving as tasks execute
• Live WebSocket connectivity indicators
• Real agent data replacing mock data

This is what convinced me the system was ready for production use.

System Architecture

I designed this with modularity and real-world use in mind:

• FastAPI Backend: Robust API server with WebSocket support for real-time updates
• React Dashboard: Clean, responsive interface for monitoring agent activity
• Linear Workflow Engine: Coordinates text analysis, data processing, and API integration agents
• WebSocket Integration: Live updates flow from backend to frontend automatically
• Redis State Management: Optional persistence (falls back to mock for development)
• Agent Orchestration: Three specialized agents working in coordinated workflows

Current Agent Types

The system includes these production-ready agents:

• Text Analyzer: Processes and analyzes text content with sentiment analysis
• Data Processor: Enriches and transforms data with statistical analysis
• API Client: Handles external API interactions and data fetching

Each agent reports status changes in real-time to the dashboard.

Why I Built This

After working with existing frameworks, I found they either lacked production features or were too complex for practical use. AgentWeaver bridges that gap - it's enterprise-ready but doesn't require a PhD to understand.

What makes this different:

• Real-time visibility: You can actually see what your agents are doing
• Works out of the box: No complex configuration or external dependencies required
• Production tested: Handles real workloads with proper error handling
• Live demos: Working examples that show the system in action

The real-time dashboard was the game-changer for me - finally being able to see agent coordination happening live made debugging and optimization so much easier.

Quick Start Demo

# Terminal 1: Start the backend
python main.py

# Terminal 2: Start the frontend  
cd frontend && npm run dev

# Terminal 3: See it in action
python examples/demos/dashboard_demo.py

Watch http://localhost:3000 and see your agents spring to life! 🚀