本文へ移動
cccskills
無料GitHub で公開

simple-modern-uv

Set up, modernize, or update Python projects using the simple-modern-uv template: uv, ruff, BasedPyright, pytest, GitHub Actions CI, and tag-driven PyPI publishing. Use when starting a new Python project or package; when modernizing, upgrading, or migrating an existing Python package to uv (from Poetry, setuptools, pip, requirements.txt, or PDM); or when updating a project already built from simple-modern-uv to the latest template.

日本語の概要は準備中です。原文の説明を表示しています。

jlevy/flowmarkスター 83(GitHub 83 + cccskills 0)
連携ツールGitHub

実行スコア 未検証

インストール方法を見る

含まれるファイル(4)

  • SKILL.md7.2 KB
  • references/adopt-existing.md6.3 KB
  • references/customize.md3.4 KB
  • references/faq.md4.0 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

simple-modern-uv: Modern Python Project Setup

simple-modern-uv is a minimal, modern Python project template: uv for everything package-related, ruff for lint/format, BasedPyright for type checking, pytest, GitHub Actions CI, and tag-driven publishing to PyPI with dynamic versioning (the git tag is the version). It is a Copier template, so projects can pull future template improvements at any time with copier update.

There are three workflows. Pick by situation:

  1. New project: create a project from scratch.
  2. Upgrade existing project: convert any existing Python package to this template’s structure and tooling. See references/adopt-existing.md.
  3. Update templated project: the project has a .copier-answers.yml from this template; pull the latest template changes.

For all workflows: when something fails, check references/faq.md before improvising. For post-setup customization (license change, private/unpublished package, entry points), see references/customize.md.

If this file was fetched from a URL rather than installed as a folder, fetch the reference files from the same ref: replace SKILL.md in the URL you used with references/<name>.md (don’t mix a pinned SKILL.md with references from main).

The Interview: What to Ask the User (and What Not to Ask)

Every workflow uses the same answer set (the template’s questions in copier.yml). Infer everything you can, then ask the user to confirm once, in a single batched message. Never ask one question at a time, and never ask about things listed under “conventions” below.

Essentials (confirm these; they are the user’s decisions):

Answer keyMeaningNotes
package_namePyPI package and GitHub repo nameNormalize to kebab-case (acme-widgets); show the user the name you derived
package_modulePython module nameDerived automatically as snake_case (acme_widgets); show it, don’t ask
package_descriptionOne-line description
package_licenseMIT (default), Apache-2.0, BSD-3-Clause, AGPL-3.0-or-later, Proprietary, or None (decide later)Always surface, naming the default
publish_to_pypitrue (default) or falseAlways surface; false for private packages and apps

For package_author_name, package_author_email, and package_github_org, infer values from git config user.name and user.email and the repo’s remote or gh auth status, and include them in the confirmation summary rather than asking.

Conventions (apply silently; mention in your final summary; deviate only if the user explicitly asks): src/ layout; first version is the git tag v0.1.0 for a new project (for upgrades of published packages, the next version above what’s on PyPI); Python 3.11+; line length 100; the template’s tool choices. These are never questions.

Workflow 1: New Project

  1. Run the interview (above). Then render, replacing the answer values:

    uvx --exclude-newer "14 days" copier@9.16.0 copy --defaults \
      --data package_name=acme-widgets \
      --data "package_description=One-line description" \
      --data "package_author_name=Jane Doe" \
      --data package_author_email=jane@example.com \
      --data package_github_org=acme \
      --data package_license=MIT \
      --data publish_to_pypi=true \
      gh:jlevy/simple-modern-uv acme-widgets
    

    (Omit --data keys that should take defaults. package_module derives automatically from package_name.)

  2. Initialize git, commit, then install. Sync only after the first commit: dynamic versioning reads git, and syncing first leaves the editable install’s version at 0.0.0.

    cd acme-widgets
    git init --initial-branch=main
    git add . && git commit -m "Initial commit from simple-modern-uv."
    make install
    git add uv.lock && git commit -m "Add uv.lock."
    
  3. Verify: make lint and make test must pass. Fix anything that doesn’t before declaring success.

  4. Offer next steps (don’t just stop): create the GitHub repo and push (an empty repo with no README, license, or .gitignore; the template provides them); fill in README.md; if publishing, docs/publishing.md covers the one-time PyPI Trusted Publisher setup; tag v0.1.0 when ready to release.

Workflow 2: Upgrade an Existing Project

Follow references/adopt-existing.md. In brief: run the interview with answers inferred from the repo itself (pyproject/setup.py metadata, LICENSE file, git remote, PyPI presence), render the template into a sibling temp directory with those answers, merge the template’s structure into the project (preserving its dependencies, metadata, and code), translate old tool configs to uv equivalents, write .copier-answers.yml so future updates work, and verify with make install, lint, and tests. Land everything on a branch as one reviewable change.

Workflow 3: Update a Templated Project

Requires a clean working tree and the project’s .copier-answers.yml.

  1. If the template has questions the project has never answered (keys present in the template’s copier.yml but missing from the project’s .copier-answers.yml), first check the project’s actual state and pass matching --data so defaults can’t revert hand customizations (see “Reconciling New Questions on Update” in references/customize.md).

  2. uvx --exclude-newer "14 days" copier@9.16.0 update --defaults --skip-answered
    
  3. Resolve any *.rej files or inline conflict markers, re-run make lint and make test, then summarize what the template update changed (check the release notes).

Verification Gate (All Workflows)

Before reporting success: make install, make lint, and make test all pass, and for publishable packages make build produces a wheel with a sensible version (requires at least one git commit; see the FAQ if the version looks wrong). Report what you ran and the results.

<!-- This document follows common-doc-guidelines.md. See github.com/jlevy/practical-prose and review guidelines before editing. -->

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

コメント

ログインなしで書き込めます。使ってみた感想、質問、動かなかった環境などを気軽にどうぞ。

まだコメントはありません。最初のコメントを書いてみませんか。

送信するとすぐに公開されます。メールアドレスや電話番号などの個人情報は書かないでください。URL は 2 つまでです。

0 / 1000 文字

このフォームは reCAPTCHA で保護されており、Google のプライバシーポリシーと利用規約が適用されます。

同じリポジトリのスキル

flowmark

無料

Fast, consistent Markdown auto-formatter for typographic cleanup, normalization, and clean semantic line breaks. Use when creating, editing, or cleaning Markdown; formatting LLM-generated docs; adopting Flowmark in a repository; adding Markdown format scripts or commit hooks; or replacing Prettier or another Markdown formatter.

日本語の概要は準備中です。原文の説明を表示しています。

jlevy/flowmark832026年9月5日 更新

tbd

無料

Git-native issue tracking (beads), coding guidelines, knowledge injection, and spec-driven planning for AI agents. Drop-in replacement for bd/Beads with simpler architecture. Use for: tracking issues/beads with dependencies, creating bugs/features/tasks, planning specs, implementing features from specs, code reviews, committing code, creating PRs, loading coding guidelines (TypeScript, Python, Rust, TDD, golden testing, Convex, monorepo patterns), code cleanup, research briefs, architecture docs, agent handoffs, external tracker and Linear integration setup, personal Linear API key onboarding, viewing beads in a live browser, and checking out third-party library source code. Invoke when user mentions: tbd, beads, bd, shortcuts, issues, bugs, tasks, features, epics, todo, tracking, specs, planning, implementation, validation, guidelines, templates, commit, PR, pull request, code review, testing, TDD, test-driven, golden testing, snapshot testing, TypeScript, Python, Rust, Convex, monorepo, cleanup, dead code, refactor, handoff, research, architecture, labels, search, web, browser, live view, external tracker, Linear, Linear API key, integration setup, checkout library, source code review, or any workflow shortcut.

日本語の概要は準備中です。原文の説明を表示しています。

jlevy/flowmark832026年9月5日 更新

jlevy のスキルをすべて見る

このスキルの問題を報告する