How yameogo123/squeleton works
Project type and purpose
• This repo is not a single runnable application but a set of “cookiecutter” skeleton templates.
– pythonPackage (general Python package)
– fastAPI (Python web-API)
– ML (machine-learning pipeline)
– reactJS (front-end SPA)
– plus an upcoming reactNative template
• Developers invoke cookiecutter on one of these directories to generate a new project, supplying name, author, etc.
File structure
• Top-level directories ML/, fastAPI/, pythonPackage/, reactJS/, reactNative/
– Each contains:
• cookiecutter.json (the user-prompt schema)
• hooks/ (pre-prompt.py or similar scripts to customize prompts)
• a subfolder named {{cookiecutter.project_name}} which holds the Jinja2-templated skeleton
• common files in each template: Dockerfile, Makefile, README.md, .gitignore, pyproject.toml or package.json, requirements*.txt, docker-compose.yaml
• Under each template’s skeleton:
– app/ (backend/frontend)
– data/ (raw/interm/processed for ML)
– docs/ (Sphinx or rst-based docs)
– notebooks/ (ML Jupyter notebooks)
– src/ (source code modules, entry-point scripts)
– tests/
Key insights from README
• Shows how to bootstrap a new project via:
cookiecutter https://github.com/Yameogo123/squeleton.git --directory="ML"
• Emphasizes prerequisites (node/npm or pip) and installing cookiecutter
a. Components
• User (developer) CLI
• Cookiecutter Engine (local or remote)
• Template Repository (this GitHub repo)
• Selected Template Directory (e.g., ML/)
• Prompt Hooks (pre_prompt.py)
• Jinja2 Template Files (inside {{cookiecutter.project_name}}/)
• Output Project (scaffolded folder tree)
b. Interactions
1. User runs cookiecutter CLI with URL+--directory.
2. Cookiecutter Engine clones or reads the repo.
3. Engine reads cookiecutter.json to gather variables.
4. Engine executes pre-prompt hooks to refine prompts.
5. Engine renders each Jinja2 file in the chosen template, substituting variables.
6. Engine writes the generated project tree to user’s disk.
c. Architectural patterns & principles
• Template-engine pattern (Jinja2 templating)
• Hook-based customization (pre-prompt scripts)
• Directory convention for consistent scaffolding
d. Relevant technologies
• Python & pip (cookiecutter, Sphinx docs)
• Node & npm (if using reactJS or reactNative templates)
• Docker & docker-compose, Makefile for containerized dev
• Jinja2, cookiecutter.json for templating