agentleFS
TsukiNi22TsukiNi22/ai-utils/skills/python/python-project/SKILL.md

python-project

Skill1 starsChanged 3 days agoMITView on GitHub

Set up a new Python project in Tsukini's way - a script, an application in the layout of his MAGIC project (src/main.py + app.py + const.py, Class/, tool/, data/, requirements.txt) or an installable package (src layout, __init__.py with the version and the public API, __main__.py, const.py with frozen dataclasses, exception.py with error codes, main() returning the exit code, pyproject.toml with a console command, pytest tests) - generated by scripts/new_project.py with the Xartania headers, then README / license / CI / git with the other skills. Use whenever the user wants to create, bootstrap or scaffold a Python project, package, tool or script project.

  • Installs packages

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    CI/CD: when the project is (or becomes) a GitHub repository (git remote -v shows github.com, or the Dépôt GitHub extra), the ci-cd skill is …

    line 23
  2. Very specific:

    Every kind gets .gitignore (venv, caches, build outputs) and a README.md stub to finish with readme-style.

    line 33
  3. Very specific:

    Check: xstyle --rtk -r . (generic + PY-* rules), fix with xstyle --rtk --fix.

    line 54

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Python project

SKILL_DIR = the directory of this file. The code follows python-style (load it). Generator:

python3 SKILL_DIR/scripts/new_project.py <name> --kind script|app|package --dir <parent> --desc "<one line>" \
    [--python 3.12] [--version 1.0.0] [--banner xartania|none|<text>]

1. Ask (one AskUserQuestion call, French) - only what the request doesn't say

  • Type: Package installable (Recommandé pour un outil / une lib) · Application (structure MAGIC) · Script seul.
  • Nom (folder, distribution and command name; the package is the name in lower case with _), description, version Python minimale (default: the installed one, python3 --version, at least 3.10 for X | None).
  • Extras (multiSelect): Tests pytest · README (readme-style) · Licence (license) · git init + premier commit (git-conventions) · Dépôt GitHub (gh repo create).
  • CI/CD: when the project is (or becomes) a GitHub repository (git remote -v shows github.com, or the Dépôt GitHub extra), the ci-cd skill is loaded after the generation: it asks what to run (tests, style, coverage, PyPI, release...) and writes the workflows (templates/python.yml, pypi.yml, release.yml).

2. Layouts

KindFilesRun
packagepyproject.toml, src/<pkg>/{__init__,__main__,const,exception,main}.py, tests/test_main.pypip install -e ".[dev]" then <name> / python -m <pkg>, pytest
app (MAGIC)src/main.py (entry, guards, exit codes), src/app.py (setup + main loop), src/const.py, src/Class/ (one class per file), src/tool/ (one function per file, named like it), data/ (runtime files), requirements.txtpython3 src/main.py
script<name>.py (##### Program ##### at module level), const.pypython3 <name>.py
Every kind gets .gitignore (venv, caches, build outputs) and a README.md stub to finish with readme-style.
  • Version: one source, VERSION = "x.y.z" at the top of const.py (a literal: setuptools reads it without importing), reused by VALUES.VERSION, __version__ and --version; bumped with the CHANGELOG (git-conventions).
  • __init__.py: the version and the public API only (__all__), no logic, no side effect at import.
  • __main__.py: exit(main()), so python -m <pkg> and the console command ([project.scripts]) are the same.
  • main.py: main(arguments: list[str] | None = None) -> int (testable without sys.argv), argparse with --version, the program in a try catching ProjectError (message on stderr, its code returned) and KeyboardInterrupt; returns RETURN.OK. Never exit() inside a package except in __main__.py.
  • exception.py: ProjectError(message, code=ERROR.ACTION, info="[None]") with formated() ([Error] message -> info, the reading of the libutils exceptions); project specific errors inherit from it.
  • const.py: the frozen dataclasses of python-style (Return, Error, Values, Files...) and their instances.
  • Dependencies: pyproject.toml dependencies (package) or requirements.txt pinned with == (app); development tools in the dev extra. A virtual environment .venv (python3 -m venv .venv), never sudo pip.
  • Data files of an application: relative to the project root through FILES in const.py, never hard-coded in the code (MAGIC's data\\parameters.save Windows paths: pathlib instead).

3. After the generation

  1. Fill const.py (names, files, values), scaffold the classes / tools / exceptions with python-class (empty bodies), then write the code with python-style and python-comments, one module per concern.
  2. Tests: tests/test_<module>.py with pytest (the tests skill for more), run them in the venv.
  3. Check: xstyle --rtk -r . (generic + PY-* rules), fix with xstyle --rtk --fix.
  4. Extras chosen: readme-style, license, git-conventions (first commit feat: initial version of <name> only if the user asked for the commit).
  5. GitHub repository (remote on github.com): ci-cd for the CI/CD (always asked there, never skipped silently).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.