cjthompson Dignified Python
Use this skill for sustained Python engineering where design quality matters. Preserve an existing repository's conventions unless the user explicitly asks to modernize them.
Establish the target
- Read the repository instructions and existing Python configuration.
- Determine the minimum Python version from
requires-python,python_requires,.python-version, CI, or equivalent evidence. - If no target is declared, ask the user. Offer Python 3.14 as the default for new work; do not silently choose a version.
- Read
dignified-python-core.md, then the matchingversions/python-3.x.mdfile.
Work deliberately
- Model domain data explicitly and keep invalid states hard to represent.
- Use precise annotations at public boundaries; avoid
Anyunless an untyped boundary truly requires it. - Prefer small cohesive modules, explicit dependencies,
pathlib, deterministic cleanup, and actionable errors. - Follow the project's established test, lint, format, type-check, build, and dependency commands.
- Make the narrowest change that solves the request. Avoid opportunistic migrations.
Load references only when needed
- CLI work:
cli-patterns.md - Subprocess work:
subprocess.md - API signatures:
references/advanced/api-design.md - Exception boundaries:
references/advanced/exception-handling.md - ABC versus Protocol:
references/advanced/interfaces.md - Casts, literals, and narrowing:
references/advanced/typing-advanced.md - Module structure and import-time behavior:
references/module-design.md - Final deep review:
references/checklists.md
For normative typing questions, hand off to python-development:python-typing-reference. For a focused annotation-tightening pass, use python-development:tighten-python-types. For tests, tooling, or concurrency, use the corresponding specialized skill in this plugin.