A minimal implementation of an AI-powered truss analysis tool that demonstrates how to combine LLMs with engineering calculations. This project shows how to create a simple agent that can interpret truss drawings and perform structural analysis.

This implementation uses Claude 3.7 to:

• Extract truss geometry from images
• Perform structural analysis using FEM
• Provide interactive analysis through a chat interface

Simply specify the path to your truss image in simple_truss_agent.py:

if __name__ == "__main__":
    # Specify the image path here - users can easily change this line
    image_path = "./images/book_truss.png"
    truss_chat(image_path) 

Run your code. The agent analyzes the image and extracts the truss geometry, including:

• Node locations
• Element connections
• Applied forces
• Support conditions
• Area (A)
• Young's modulus (E)

Chat with the agent to:

• Run structural analysis
• Modify the truss geometry
• Get detailed results
• Ask engineering questions

1. Install dependencies:

   pip install -r requirements.txt

2. Configure your Anthropic API key in config.py:

   ANTHROPIC_API_KEY = "your-api-key-here"

3. Run the application:

   python simple_truss_agent.py

The agent can:

• Analyze truss images and extract geometry
• Calculate displacements and stresses
• Answer questions about the analysis results

Example commands:

> Run analysis with E=200e9 Pa and A=0.001 m²
> Increase the height of the truss by 2 meters
> Find the smallest IPE profile we can use, while keeping the stress under 200 MPa

• simple_truss_agent.py: Main application with chat interface
• tools.py: Core functionality for truss analysis
• config.py: Configuration settings
• images/: Example truss structure images

• Clear truss image

MIT License - see LICENSE file for details.