RinDig/Interpretable-Context-Methodology ? reverse-engineered prompt

Build me a simple Python project that shows the Interpretable Context Method in action.

I want a workspace based agent workflow where the folder structure tells a single AI agent what to do at each step, instead of using a big orchestration framework. The app should create or manage numbered stage folders, with markdown files for the instructions, the context, and the outputs. Each stage should read only the files it needs, write its result into its output folder, and let a human edit those files before the next stage runs.

Please make it easy to set up a new workspace with a questionnaire, shared config for style or brand rules, and clear stage contracts that say what each stage reads, does, and writes. It should feel like a practical template for research, writing, and production style workflows, with plain text files as the main interface.

Keep it clean and simple, and if you need current Python docs or best practices, look them up online.