Agent Skills: Code Improvement & Cleanup Skill

Apply systematic improvements to code quality, performance, maintainability, and cleanup. Use when refactoring code, optimizing performance, removing dead code, or improving project structure.

UncategorizedID: tony363/superclaude/sc-improve

Install this agent skill to your local

pnpm dlx add-skill https://github.com/Tony363/SuperClaude/tree/HEAD/.claude/skills/sc-improve

Skill Files

Browse the full folder contents for sc-improve.

Download Skill

Loading file tree…

.claude/skills/sc-improve/SKILL.md

Skill Metadata

Name
sc-improve
Description
Apply systematic improvements to code quality, performance, maintainability, and cleanup. Use when refactoring code, optimizing performance, removing dead code, or improving project structure.

Code Improvement & Cleanup Skill

Systematic improvements with multi-persona expertise and safety validation.

Quick Start

# Quality improvement
/sc:improve src/ --type quality --safe

# Performance optimization
/sc:improve api-endpoints --type performance

# Dead code cleanup
/sc:improve src/ --cleanup --type code --safe

# Import optimization
/sc:improve --cleanup --type imports

Behavioral Flow

  1. Analyze - Examine codebase for improvement opportunities
  2. Plan - Choose approach and activate relevant personas
  3. Execute - Apply systematic improvements
  4. Validate - Ensure functionality preservation
  5. Document - Generate improvement summary

Flags

| Flag | Type | Default | Description | |------|------|---------|-------------| | --type | string | quality | quality, performance, maintainability, style, principles, code, imports, files, all | | --cleanup | bool | false | Enable cleanup mode | | --safe | bool | true | Conservative with safety validation | | --aggressive | bool | false | Thorough cleanup (use with caution) | | --preview | bool | false | Show changes without applying | | --interactive | bool | false | Guided decision mode |

Personas Activated

  • architect - Structure and design improvements
  • performance - Optimization expertise
  • quality - Code quality and maintainability
  • security - Security pattern application
  • code-warden - KISS and Purity enforcement (with --type principles)

MCP Integration

PAL MCP (Validation & Analysis)

| Tool | When to Use | Purpose | |------|-------------|---------| | mcp__pal__consensus | Complex refactors | Multi-model validation before major changes | | mcp__pal__codereview | Quality assessment | Review improvement quality and safety | | mcp__pal__thinkdeep | Architecture changes | Deep analysis of structural improvements | | mcp__pal__precommit | Before commit | Validate all changes preserve functionality | | mcp__pal__debug | Regression issues | Root cause analysis if improvements break things | | mcp__pal__challenge | Aggressive mode | Critical evaluation of aggressive cleanup decisions |

PAL Usage Patterns

# Consensus for major refactor
mcp__pal__consensus(
    models=[
        {"model": "gpt-5.2", "stance": "for"},
        {"model": "gemini-3-pro", "stance": "against"}
    ],
    step="Evaluate: Should we extract this into a separate module?"
)

# Review after improvements
mcp__pal__codereview(
    review_type="full",
    step="Reviewing code improvements",
    findings="Quality, maintainability, backwards compatibility",
    relevant_files=["/src/refactored/module.py"]
)

# Pre-commit validation
mcp__pal__precommit(
    path="/path/to/repo",
    step="Validating refactoring changes",
    confidence="high"
)

Rube MCP (Automation & Tracking)

| Tool | When to Use | Purpose | |------|-------------|---------| | mcp__rube__RUBE_SEARCH_TOOLS | External tools | Find linters, formatters, analyzers | | mcp__rube__RUBE_MULTI_EXECUTE_TOOL | Notifications | Update tickets, notify team | | mcp__rube__RUBE_CREATE_UPDATE_RECIPE | Reusable workflows | Save improvement patterns |

Rube Usage Patterns

# Notify team of improvements
mcp__rube__RUBE_MULTI_EXECUTE_TOOL(tools=[
    {"tool_slug": "SLACK_SEND_MESSAGE", "arguments": {
        "channel": "#refactoring",
        "text": "Completed: Dead code cleanup removed 500 lines"
    }},
    {"tool_slug": "JIRA_UPDATE_ISSUE", "arguments": {
        "issue_key": "TECH-456",
        "status": "Done"
    }}
])

# Create improvement report in Notion
mcp__rube__RUBE_MULTI_EXECUTE_TOOL(tools=[
    {"tool_slug": "NOTION_CREATE_PAGE", "arguments": {
        "title": "Refactoring Report - Q4 2025",
        "content": "## Summary\n- Lines removed: 500\n- Complexity reduced: 25%"
    }}
])

Evidence Requirements

This skill requires evidence. You MUST:

  • Show before/after code comparisons
  • Run tests to verify functionality preservation
  • Report metrics (lines removed, complexity reduction)

Improvement Types

Quality (--type quality)

  • Technical debt reduction
  • Code structure improvements
  • Readability enhancements

Performance (--type performance)

  • Bottleneck resolution
  • Algorithm optimization
  • Resource efficiency

Maintainability (--type maintainability)

  • Complexity reduction
  • Documentation improvements
  • Modular restructuring

Style (--type style)

  • Formatting consistency
  • Naming conventions
  • Pattern alignment

Principles (--type principles)

  • KISS compliance improvements (reduce complexity, extract methods)
  • Purity enforcement (separate I/O from logic)
  • Guard clause refactoring (reduce nesting)
  • Functional core extraction (move I/O to shell layer)

Validators:

# Run KISS validation
python .claude/skills/sc-principles/scripts/validate_kiss.py --scope-root . --json

# Run Purity validation
python .claude/skills/sc-principles/scripts/validate_purity.py --scope-root . --json

Cleanup Mode (--cleanup)

Code Cleanup (--type code)

  • Dead code detection and removal
  • Unused variable elimination
  • Unreachable code removal

Import Cleanup (--type imports)

  • Unused import removal
  • Import organization
  • Dependency optimization

File Cleanup (--type files)

  • Empty file removal
  • Orphaned file detection
  • Structure optimization

Full Cleanup (--type all)

  • Comprehensive cleanup
  • All categories combined
  • Multi-persona coordination

Safety Modes

Safe Mode (--safe)

  • Conservative changes only
  • Automatic safety validation
  • Preserves all functionality

Aggressive Mode (--aggressive)

  • Thorough cleanup
  • Framework-aware patterns
  • Requires careful review

Examples

Safe Quality Improvement

/sc:improve src/ --type quality --safe
# Technical debt reduction with safety validation

Performance Optimization

/sc:improve api-endpoints --type performance --interactive
# Guided optimization with profiling analysis

Dead Code Cleanup

/sc:improve src/ --cleanup --type code --safe
# Remove unused code with dependency validation

Preview Changes

/sc:improve --cleanup --type imports --preview
# Show what would be removed without executing

Tool Coordination

  • Read/Grep/Glob - Code analysis
  • Edit/MultiEdit - Safe modifications
  • TodoWrite - Progress tracking
  • Task - Large-scale improvement delegation