In this proof of concept project we attempt to teach ChatGPT basic UML to generate code, consume the output and create actual files in a project. The idea is for developers to provide a schema of sorts and for ChatGPT to generate some boilerplate code to avoid set up time.

The goal is to go from this prompt

+ sign in fragment (SIF)
+ sign up fragment (SUF)
+ authentication viewmodel (AVM)
+ authentication repository (AR)
+ user repository (UR)
- navigation manager (NM)
+ authentication service (AS)
SIF --> AVM
SUF --> AVM
AVM --> AR
AVM --> UR
AVM --> NM
AR --> AS

to these files

Project was tested with python3

pip install openai
pip install openai[datalib]
pip install tiktoken
pip install halo

Create/Get your ChatGPT api token here and use it here.

openai.api_key = "<API_TOKEN>"

Edit your training sample files here

Edit the diagram here

Run any script with python

python chat-gpt-uml-kotlin.py --package auth --diagram diagram --samples samples --debug