An intelligent agent built with the Claude Agent SDK that analyzes blockchain security incidents and generates interactive 3D visualizations of transaction networks.

This agent automatically:

1. Retrieves blockchain addresses involved in security incidents
2. Fetches structured metadata using TRM Labs BlockInt API
3. Analyzes transaction networks and risk assessments
4. Generates interactive three.js force-directed graph visualizations
5. Creates comprehensive HTML reports with technical summaries

• Autonomous Analysis: Claude-powered agent orchestrates the entire workflow
• Risk Visualization: Node colors represent risk scores (red=high, orange=medium, green=low)
• Network Mapping: Force-directed graph shows relationships between entities
• Interactive 3D: Rotate, zoom, and click nodes for detailed information
• Transaction Flow: Edge thickness indicates transaction volume

1. Clone the repository:

git clone https://github.com/jknoll/agent-blockchain-visualization.git
cd agent-blockchain-visualization

2. Create a virtual environment and install dependencies:

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Set up environment variables:

cp .env.example .env
# Edit .env and add your API keys:
# ANTHROPIC_API_KEY=your_anthropic_key
# TRM_API_KEY=your_trm_key (optional - uses mock data if not provided)

Run the agent:

python run.py

The agent will:

• Load incident data from data/addresses.json
• Query the TRM Labs API (or use mock data)
• Generate visualization at output/[incident-id]/incident.html
• Provide a detailed analysis summary

agent-blockchain-visualization/
├── src/
│   ├── agent.py              # Main agent orchestration
│   ├── data_retrieval.py     # Data loading from local/remote sources
│   ├── trm_api.py           # TRM Labs API integration
│   └── visualization.py      # Three.js HTML generation
├── data/
│   └── addresses.json        # Incident data store
├── output/                   # Generated visualizations
├── requirements.txt
├── run.py                    # Entry point
└── README.md

Edit data/addresses.json to add new incidents:

{
  "incidents": [
    {
      "id": "incident-001",
      "name": "Security Incident Name",
      "blockchain": "ethereum",
      "addresses": ["0x..."],
      "transaction_ids": [],
      "description": "Incident description"
    }
  ]
}

• ANTHROPIC_API_KEY (required): Get from https://console.anthropic.com/
• TRM_API_KEY (optional): Get from https://www.trmlabs.com/
  • Without TRM API key, the system uses realistic mock data for testing

The agent uses Claude's tool-calling capabilities to:

1. get_incident_data: Load incident from datastore
2. fetch_address_info: Query TRM Labs for address details
3. fetch_risk_assessment: Get risk scores and indicators
4. fetch_transaction_network: Build network graph
5. generate_visualization: Create interactive HTML

The agent autonomously decides which tools to use and in what order, adapting to the data it receives.

• Left click + drag: Rotate the view
• Right click + drag: Pan
• Scroll: Zoom in/out
• Click nodes: Display detailed information
• Reset View button: Return to default camera position
• Pause/Play button: Toggle force simulation

# Test data retrieval
python src/data_retrieval.py

# Test TRM API
python src/trm_api.py

# Test visualization
python src/visualization.py

Add tool definitions in src/agent.py and implement handlers in process_tool_call().

• Python 3.9+
• Anthropic API key
• TRM Labs API key (optional)

ISC

See milestones.md for development roadmap.