cleanup-code

Code Cleanup

Perform cleanup-focused analysis and refactoring of the target code.

Step 1: Detect Language

From $ARGUMENTS (target path) or git diff HEAD, determine which languages are in scope:

• .rs files or Cargo.toml → Rust
• .ts/.tsx files or package.json → TypeScript
• .py files or pyproject.toml/setup.py → Python

Step 2: Load Standards

Read ONLY the matching standards files from references/ (relative to SKILL.md):

• Rust → references/standards-rust.md
• TypeScript → references/standards-typescript.md
• Python → references/standards-python.md

Also read ${CLAUDE_PLUGIN_ROOT}/rules/planning.md for design principles.

Step 3: Cleanup Analysis

Review the target code looking for cleanup opportunities. Think like a PR reviewer focused on code quality.

Focus areas (all languages):

• Dead code and unused imports
• Code duplication (DRY violations)
• Files exceeding 800 lines — split them; use code-simplifier and pr-review-toolkit plugins
• Overly complex functions (candidates for simplification)
• Law of Demeter violations (excessive chaining, inappropriate coupling)
• YAGNI violations (unused abstractions, over-engineering)

Language-specific focus (from loaded standards):

• Apply the error handling, naming, and organization rules from the loaded standards file
• Check for deprecated API usage per the standards

Step 4: Relationship Mapping

Use ast-grep for structural pattern matching first, then serena's symbolic tools (find_symbol, find_referencing_symbols, get_symbols_overview) to understand code relationships before proposing changes. Identify methods that can be consolidated and reused rather than duplicated.

Step 5: Refactoring Plan

Generate a structured refactoring plan following the design principles from planning.md. Prioritize changes by impact and risk. Group related changes for atomic commits.

Step 6: Review Gate

Present the refactoring plan for approval. Clearly identify any breaking changes or risks.