cryptocurrency-trading-ai-agent-agishark

🦈 Cryptocurrency Trading AI Agent [AGISHARK]

📌 Overview

This project was developed as part of the AGI Agent Application Hackathon. It's an AI-based trading bot that can check cryptocurrency market information, analyze trading strategies, and execute real trades.

🚀 Key Features

• ✅ Powerful AI Agent:
  • Web search capability
  • X (Twitter) search
  • Autonomous order execution and listing
  • Document database access
  • Multiple AI model selection options
• ✅ Real-time Exchange Information with Upbit API:
  • Price and chart viewing for major cryptocurrencies
  • Market trend analysis and visualization
  • Asset management
  • Coin transaction history
• ✅ Automated Investment Strategies:
  • AI Agent's automated trading system
  • Customized investment instructions and preference management
  • Large-volume investment guidance management through PDF and RAG documents

🖼️ Demo / Screenshots

👆 Click on the image above to watch the demo video

🧩 Tech Stack

• Frontend: Streamlit
• Backend: Python
• Database: JSON for state management
• Others: OpenAI API, Upstage API, Upbit API, X Bearer Token, Vector stores for RAG

🏗️ Project Structure

📁 cryptocurrency-trading-ai-agent-agishark/
├── app.py                    # Main application entry point
├── init.py                   # Initialization file
├── requirements.txt          # Python dependencies
├── .gitignore                # Git ignore file
├── page/                     # Streamlit pages
│   ├── api_setting.py        # API configuration interface
│   ├── auto_trader_page.py   # Automated trading interface
│   ├── portfolio.py          # Portfolio management
│   ├── sidebar.py            # App sidebar component
│   ├── trade_history.py      # Trading history view
│   ├── trade_market.py       # Market view
│   └── trade_strategy.py     # Strategy configuration
├── tools/                    # Core functionality tools
│   ├── auto_trader/          # Automated trading logic
│   ├── document_parser/      # PDF and document processing
│   ├── information_extract/  # Data extraction utilities
│   ├── rag/                  # Retrieval-Augmented Generation
│   ├── search_X/             # X (Twitter) search functionality
│   ├── upbit/                # Upbit API integration
│   ├── web2pdf/              # Web to PDF conversion
│   └── web_search/           # Web search capabilities
├── model/                    # AI models
│   └── open_ai_agent.py      # OpenAI integration
├── util/                     # Utility functions
└── data/                     # Data storage
    ├── api_key_store.json    # API key storage
    ├── agent_state.json      # Agent state management
    ├── agent_work_time.json  # Agent work time tracking
    └── vector_store_id.json  # Vector storage IDs

🔧 Setup & Installation

cd cryptocurrency-trading-ai-agent-agishark/code_kor

# Install required packages
pip install -r requirements.txt

# Run the application
streamlit run app.py

📁 Required APIs & References

• Required APIs:
  1. Upstage API Key (Document Parser, Information Extracter)
  2. OpenAI API Key (OpenAI Agent, Web Search, Vector Store)
     • Available at: https://platform.openai.com/api-keys
  3. Upbit Access Key, Secret Key
     • Requires K-Bank account setup
     • Available at: https://upbit.com/mypage/open_api_management
     • Note: One key can be used from only one IP address at a time
  4. X Bearer Token (Search)
     • Login at: https://developer.x.com/en/portal/dashboard
     • Click the key icon in Project App
     • Generate Bearer Token from Authentication Tokens

🙌 Team Members

Name	Role	GitHub
Jaewan Shin	Development Lead, Frontend, Agent Design, RAG	@alemem64
Jihun Jang	Investment Strategy Page, Document Management System, Web2PDF, Presentation	@pinesound05
Yujin Cha	Web Search, X Search, Document Parser, Information Extract Integration	@yujinc726
Mingyu Shin	Upbit Agent Tool Development	@girafxxx-beep

⏰ Development Period

• 2025-03-29:

  • Initial project setup and repository creation

• 2025-03-30 ~ 2025-03-31:

  • Basic UI construction (Streamlit)
  • Upbit API integration (account connection, trading functionality)
  • Document Parser and Information Extract testing
  • Web-to-PDF conversion feature implementation
  • Sidebar and chat interface development

• 2025-04-01 ~ 2025-04-02:

  • OpenAI Agent integration and conversation history functionality
  • Agent restart feature implementation
  • Investment strategy page UI and functionality (upload/download/delete)
  • Investment preference settings (customized instructions, risk tolerance, trading period)
  • Web2PDF features and document management system enhancement
  • X (Twitter) search tool addition

• 2025-04-03 ~ 2025-04-04:

  • AI Agent tools expansion (trading coin list, price lookup, trade execution)
  • WebSearch tool and X (Twitter) search completion
  • RAG (Retrieval-Augmented Generation) integration
  • Automated trading system development and UI improvements
  • Exception handling and bug fixes
  • README documentation and final refinements

📄 License

This project is licensed under the MIT license.
See the LICENSE file for more details.

💬 Additional Notes

• Recommended Environment: Python 3.11.11

• Troubleshooting:

  1. For numpy/pandas installation errors:

     pip uninstall numpy pandas
     pip install numpy>=1.26.0 pandas>=2.1.0

  2. For Mac users: Run this before installation:

     pip install --upgrade pip wheel setuptools