AgentSkillsCN

python-refactor

协助进行自动化、安全且可解释的Python重构。它能自动检测常见的反模式，并基于抽象语法树（AST）提供转换方案，同时以补丁形式进行修改，且改动易于单元测试。

SKILL.md
--- frontmatter
name: python-refactor
description: Assist with automated, safe, and explainable Python refactorings. Detects simple anti-patterns and offers AST-based transformations with patches and unit-test friendly changes.

Python Refactor Skill

Overview

This skill helps authors and agents perform automated, small, safe Python refactorings. It focuses on detectable, low-risk transformations (formatting to f-strings, simplify boolean expressions, remove unused imports, small function extraction suggestions) and produces patch files or dry-run diffs.

Use this skill when you want a reproducible, reviewable set of refactor suggestions that can be applied automatically or passed to a human reviewer.

Quick Start

  • •Use the scripts/analyze.py to scan a directory for refactor opportunities.
  • •Use the scripts/refactor.py to generate patches (preview mode by default).
  • •Use the scripts/apply_patch.py to apply confirmed patches safely (creates backups).

When to use

  • •Large repositories where repetitive low-risk refactors are needed.
  • •Codebases with inconsistent formatting or migration to modern Python features (f-strings, type hints).
  • •As a pre-PR pass to reduce mechanical review comments.

Scripts

  • •scripts/analyze.py: static analysis for refactor candidates.
  • •scripts/refactor.py: performs AST transforms and outputs unified diffs.
  • •scripts/apply_patch.py: applies diffs to files with backups.

Resources

  • •references/python-style-guide.md: opinionated style guidance used by the skill.
  • •references/refactoring-patterns.md: patterns the skill recognizes and how it changes code.

Safety and Scope

This skill only performs transformations that can be verified via unit tests. Always run tests after applying patches and review diffs before committing.