UML/Mermaid to Code Generation Experiment

Overview

This project demonstrates automated code generation from UML diagrams using an LLM (Claude Opus 4.1).

Objective

Explore how diagram specifications translate into working code implementations via LLM interpretation.

Key Observations

• LLM inferred security practices (password hashing, secure tokens)
• Added practical features not in diagrams (session expiry, validation)
• Maintained language-appropriate patterns (Python properties, JS prototypes)
• Preserved diagram relationships in code structure

Generation Details

• Model: Claude Opus 4.1
• Tool: Claude Code
• Prompt Used: "generate the associated code for any plantuml or mermaid files. Put the code in the same directory as the uml or mermaid file it is associated with."
• Method: Direct diagram interpretation to idiomatic code from single prompt