Skip to content

CEDAR_MASTRA_SETUP.md

Latest commit

 

History

History
159 lines (122 loc) · 4.46 KB

CEDAR_MASTRA_SETUP.md

File metadata and controls

159 lines (122 loc) · 4.46 KB

Cedar OS + Mastra Backend Setup

This project now uses Cedar OS frontend with a Mastra backend for intelligent research paper analysis.

🏗️ Architecture

  • Frontend: Cedar OS (React) - Enhanced chat interface
  • Backend: Mastra - AI agent orchestration
  • AI: OpenAI GPT-4o - Research analysis

🚀 Quick Start

1. Set Environment Variables

export OPENAI_API_KEY='your-openai-api-key-here'

2. Start Both Services

./start_cedar_mastra.sh

This will start:

  • Mastra backend on http://localhost:4111
  • React frontend on http://localhost:3000

3. Test the Backend

node test_mastra_backend.js

🤖 How It Works

Mastra Backend (mastra-backend/)

  • Research Analysis Agent: Analyzes papers and provides insights
  • Context Awareness: Receives selected papers and graph data
  • Intelligent Responses: Generates comprehensive analysis

Cedar OS Frontend (refnet/frontend/)

  • FloatingCedarChat: Enhanced chat interface
  • State Management: Tracks selected papers and graph context
  • Real-time Communication: Connects to Mastra backend

📊 Features

Research Analysis

  • Single Paper Analysis: Deep dive into individual papers
  • Comparative Analysis: Compare multiple papers
  • Pattern Recognition: Find common themes and relationships
  • Gap Identification: Discover research opportunities

Smart Context

  • Paper Selection: Automatically receives selected papers
  • Graph Data: Understands paper relationships
  • Dynamic Responses: Adapts to different query types

🔧 Configuration

Mastra Backend

  • Port: 4111 (configurable in server.js)
  • CORS: Enabled for localhost:3000 and localhost:3001
  • Model: GPT-4o for high-quality analysis

Cedar OS Frontend

📝 Usage

  1. Select Papers: Click on papers in the graph interface
  2. Open Chat: Click the chat button to open FloatingCedarChat
  3. Ask Questions: Type questions about the selected papers
  4. Get Insights: Receive intelligent analysis and comparisons

Example Queries

  • "What do you know about these papers?"
  • "Compare the methodologies used"
  • "What are the main findings?"
  • "Identify research gaps"
  • "How do these papers relate to each other?"

🛠️ Development

Backend Development

cd mastra-backend
npm run dev  # Auto-restart on changes

Frontend Development

cd refnet/frontend
npm start  # Hot reload enabled

Testing

# Test Mastra backend
node test_mastra_backend.js

# Test full integration
./start_cedar_mastra.sh

📁 File Structure

RefNet/
├── mastra-backend/           # Mastra backend server
│   ├── server.js            # Main server with research agent
│   └── package.json         # Backend dependencies
├── refnet/frontend/         # Cedar OS frontend
│   ├── src/components/
│   │   └── FloatingCedarChat.js  # Enhanced chat component
│   └── src/cedar/           # Cedar configuration
├── start_cedar_mastra.sh    # Startup script
└── test_mastra_backend.js   # Backend test script

🔍 Troubleshooting

Backend Issues

  • Check if Mastra backend is running on port 4111
  • Verify OpenAI API key is set
  • Check console logs for errors

Frontend Issues

  • Ensure backend is running before starting frontend
  • Check browser console for connection errors
  • Verify CORS settings in backend

Connection Issues

🎯 Benefits

Over Custom Python Backend

  • ✅ Pure Cedar OS: Uses official Cedar + Mastra stack
  • ✅ Less Code: Simpler implementation
  • ✅ Better Integration: Native Cedar OS features
  • ✅ Easier Maintenance: Standard framework patterns

Research Analysis Features

  • ✅ Intelligent Analysis: GPT-4o powered insights
  • ✅ Context Awareness: Understands paper relationships
  • ✅ Adaptive Responses: Different analysis types
  • ✅ Rich Output: Formatted, structured responses

🚀 Next Steps

  1. Start the services: ./start_cedar_mastra.sh
  2. Test the integration: Select papers and ask questions
  3. Customize agents: Modify mastra-backend/server.js
  4. Add features: Extend the research analysis capabilities

The system is now ready for intelligent research paper analysis! 🎉