This document describes the multi-agent architecture for the AI-powered cryptocurrency trading system.
The system is organized into four specialized agents, each responsible for different aspects of the trading platform:
- Frontend Developer - User interface and real-time visualization
- Backend Developer - API, database, and system infrastructure
- Trading Systems Developer - Trading strategies and execution
- AI/ML Developer - Predictive models and machine learning
- Model: claude-opus-4-1-20250805
- Specialization: React, TypeScript, UI/UX, WebSocket integration
- Workspace:
src/frontend/
- Context Files:
src/frontend/**
, package.json
Responsibilities:
- Real-time trading interface development
- WebSocket integration for live market data
- Performance optimization
- Responsive design implementation
Documentation: src/frontend/README.md
- Model: claude-opus-4-1-20250805
- Specialization: Node.js, Python, APIs, Database design
- Workspace:
src/backend/
- Context Files:
src/backend/**
, requirements.txt
Responsibilities:
- RESTful API development
- Database schema design and optimization
- Security implementation
- External exchange integration
- Error handling and validation
Documentation: src/backend/README.md
- Model: claude-opus-4-1-20250805
- Specialization: Trading algorithms, Market analysis, Risk management
- Workspace:
src/trading/
, strategies/
- Context Files:
src/trading/**
, strategies/**
Responsibilities:
- Trading strategy implementation
- Technical indicator development
- Risk management systems
- Backtesting framework
- Order execution logic
Documentation: src/trading/README.md
- Model: claude-opus-4-1-20250805
- Specialization: Machine Learning, Deep Learning, Predictive modeling
- Workspace:
src/ml/
, models/
- Context Files:
src/ml/**
, models/**
Responsibilities:
- Feature engineering
- Model development and training
- Real-time inference
- Model evaluation and monitoring
- Integration with trading systems
Documentation: src/ml/README.md
Market Data → Backend → AI/ML Agent → Trading Agent → Backend → Frontend
↓ ↓
Database Execution
-
Backend ↔ Frontend
- REST APIs for data retrieval
- WebSocket for real-time updates
- Authentication and session management
-
Backend ↔ AI/ML
- Feature data pipelines
- Model inference endpoints
- Performance metrics collection
-
AI/ML ↔ Trading
- Prediction signals
- Market sentiment analysis
- Risk score calculation
-
Trading ↔ Backend
- Order execution requests
- Position management
- Trade history storage
Agent configuration is defined in
agents-config.yaml
. To modify agent settings:
- Edit the YAML configuration file
- Update relevant workspace documentation
- Reinitialize affected agent workspaces if needed
Each agent workspace contains:
README.md
- Agent-specific documentation
- Directory structure for organized development
.gitkeep
files to preserve empty directories
To work with a specific agent:
- Navigate to the agent's workspace directory
- Review the README.md for responsibilities and guidelines
- Follow the directory structure conventions
- Ensure code adheres to agent specialization
-
Planning Phase
- Identify which agent(s) should handle the task
- Review agent responsibilities and context
-
Implementation Phase
- Work within the designated agent workspace
- Follow agent-specific guidelines
- Coordinate with other agents at integration points
-
Testing Phase
- Unit tests within agent workspace
- Integration tests across agent boundaries
- End-to-end testing through the full pipeline
-
Deployment Phase
- Deploy backend and ML services
- Deploy frontend to hosting platform
- Monitor agent-specific metrics
- Keep agent workspaces isolated to maintain clear boundaries
- Use shared types/interfaces for cross-agent communication
- Document integration points clearly
- Follow agent-specific coding conventions
- Maintain comprehensive test coverage within each workspace
Initialization Status: ✓ Complete
All agent workspaces have been initialized and are ready for development.
Created: 2025-10-30
Last Updated: 2025-10-30