Module Scaffolder

module-scaffolder

Sets up the skeleton for a new code project. Triggers when the user starts a new project - whatever the type.

When this skill triggers

When the user creates a new folder for a project or says "I'm starting a new project". Run it before the first code, not after.

Four project types

Each type has its own skeleton. Ask the user which type it is first, if it isn't clear from the context.

1. Python module - full standard for a reusable library
2. CLI tool - command-line tool with an entry point
3. API service - web service with endpoints and secrets
4. Learning script (own exercises) - minimal

1. Python module

Full standard. When the user builds a reusable library or a production module.

Directory structure:

module-name/
├── module_name/
│   ├── __init__.py
│   └── main.py
├── tests/
│   └── __init__.py
├── .env.example
├── .gitignore
├── README.md
├── requirements.txt
└── LICENSE

Required:

• .gitignore must contain .env
• README.md in English with the standard sections (Description, Setup, Usage, Known, Limits)
• Logging setup in __init__.py as a pattern
• First commit: feat: initial scaffold

2. CLI tool

Command-line tool. When the project should be run via python -m tool_name or a console command.

Directory structure:

tool-name/
├── tool_name/
│   ├── __init__.py
│   ├── __main__.py
│   └── cli.py
├── tests/
│   └── __init__.py
├── .gitignore
├── README.md
├── requirements.txt
└── LICENSE

Required:

• __main__.py as the entry point (python -m tool_name)
• cli.py parses arguments (argparse), one function per command
• README.md with a command-line usage example
• .gitignore with the Python standard

3. API service

Web service. When the project exposes endpoints and needs secrets or a port.

Directory structure:

service-name/
├── service_name/
│   ├── __init__.py
│   ├── main.py
│   └── routes.py
├── tests/
│   └── __init__.py
├── .env.example
├── .gitignore
├── README.md
├── requirements.txt
└── LICENSE

Required:

• .env for secrets and the port, never in the repo
• main.py starts the server
• routes.py keeps the endpoints separate from the server configuration
• README.md with a list of the endpoints
• .gitignore must contain .env

4. Learning script (own exercises)

Minimal. When working through book examples or writing small learning scripts of your own.

Directory structure:

project-name/
├── project_name.py
└── README.md

Required:

• Just a .py file plus an optional README
• No requirements.txt needed if it's stdlib only
• No git init needed if it stays purely local
• If you do use git: .gitignore with the Python standard

Templates

.gitignore (Python standard, all project types)

# Environment
.env
.env.local
.venv/
venv/

# Python
__pycache__/
*.py[cod]
*$py.class
*.so
.Python
build/
dist/
*.egg-info/

# IDE
.vscode/
.idea/
*.swp

# OS
.DS_Store
Thumbs.db

# Testing
.pytest_cache/
.coverage
htmlcov/

.env.example (module/service with API access)

# LLM API (if used)
ANTHROPIC_API_KEY=

# Local LLM (if used)
OLLAMA_HOST=http://localhost:11434

# Database (if relevant)
DATABASE_URL=

# Service (if relevant)
PORT=8000

Logging setup (init.py)

import logging

logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)

handler = logging.StreamHandler()
formatter = logging.Formatter(
    "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
)
handler.setFormatter(formatter)
logger.addHandler(handler)

requirements.txt (standard stack)

# LLM
anthropic>=0.40.0

# HTTP
requests>=2.31.0

# Environment
python-dotenv>=1.0.0

Pin exact versions if the module needs to run reproducibly.

Gotchas

Common mistakes when scaffolding - avoid them.

Committing .env into the repo

On the first commit after scaffolding: check that .env is really in .gitignore before git add . runs. Standard order: create .gitignore first, then the other files.

Folder names with spaces

Spaces in folder names break many tools (gh CLI, pip, some Python modules), especially on Windows and with cloud-sync paths. Always use a hyphen or underscore, never a space.

Module name vs. directory name

The Python module name (inner folder) needs an underscore (my_module), the repo name can have a hyphen (my-module). Keep both in mind when scaffolding — directory with a hyphen, inner module with an underscore.

Unpinned requirements.txt

A requirements.txt without version pinning (anthropic instead of anthropic>=0.40.0) breaks later when the library introduces a breaking change. At least >= with the current version, better an exact == for production modules.

A German-language README on a public repo

Public repos need an English README. If you naturally think in German, write the README in English from the start when scaffolding instead of translating it later.

Wrong location

Code belongs in the project directory, not in a notes or docs folder. Notes folders are for notes, not for code — keep the two cleanly separate.

Forgetting the LICENSE

Public modules need a LICENSE file (e.g. MIT). Without a license nobody may legally use the code. Add it right away when scaffolding, not "later".

The first commit message

First commit after scaffolding: feat: initial scaffold (Conventional Commits). Not "first commit" or "init".

Example run

User: "I'm starting a new Python module, it should become an HTTP client wrapper."

Claude Code:

1. Determines the project type → Python module (clear from the context)
2. Asks for the module name → User: "http-client-wrapper"
3. Sets up the directory structure:
   • Repo folder: http-client-wrapper/ (with a hyphen)
   • Inner module: http_client_wrapper/ (with an underscore)
4. Creates all required files from the templates
5. Writes .gitignore first, with .env in it
6. Creates the README in English with the standard sections
7. Adds the logging pattern in __init__.py
8. git init, then the first commit: feat: initial scaffold
9. Reports what was created, doesn't run ahead on its own

The user reviews it, corrects if needed, takes it over.

Source: andy-builds-ai/claude-code-skills — distributed by TomeVault.