agentsclimarketplace

Refactor

Skill Pamacea/smite/plugins/studio/skills/refactor

MANDATORY gate before ANY refactoring task in smite project. Invoke FIRST when cleaning up code, improving structure, or optimizing - provides systematic validation with 6 modes (--quick, --full, --analyze, --review, --resolve, --verify) plus 3 specialized modes (--profile for performance, --security for vulnerabilities, --types for TypeScript) with auto-team activation for complex refactors. Specific phrases: 'refactor this', 'clean up code', 'improve this function', 'optimize this', 'restructure this module'. (user)From its SKILL.md

Install
npx -y skills add Pamacea/smite --skill refactor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

6.0 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it

Refactor Skill

Mission

Provide unified, systematic code refactoring through comprehensive validation, ensuring safe improvements while preserving functionality.

What this skill accomplishes:

  • Cleans up code while maintaining functionality
  • Improves structure and maintainability
  • Optimizes performance with measured improvements
  • Enhances security through vulnerability scanning
  • Strengthens type safety in TypeScript codebases

Scope Definition

This skill applies to:

  • Code cleanup and simplification
  • Structure improvement and reorganization
  • Performance optimization with metrics
  • Security vulnerability remediation
  • Type safety enhancement
  • Duplication elimination

Related Skills

For tasks outside this scope, consider:

  • /studio build - New feature implementation
  • /studio review - Code review without refactoring
  • Edit tool - Simple renames or documentation updates
  • /studio build --test - Test writing

Tool Selection Standards

TaskRecommended ToolAvoid
Find code by meaninggrepai searchgrep/egrep/find
Read file contentsRead toolcat/head/tail
List filesGlob toolls/dir/find
Search file contentsGrep toolgrep/rg

Decision Guide:

  • Need to search? → grepai search or /toolkit search
  • Need to read? → Read tool
  • Need to list? → Glob tool
  • Need to grep? → Grep tool

Execution Protocol

Step 1: Analysis (Always First)

Understand before acting:

  1. Use semantic search to locate relevant code
  2. Read files with the Read tool
  3. Identify complexity hotspots
  4. Detect duplication patterns
  5. Assess test coverage

Step 2: Planning (For Multi-File Changes)

Create a plan when:

  • Modifying 3+ files
  • Restructuring a module
  • Running subagents
  • Using analyze/full/resolve modes

Step 3: Execution (Incremental)

Apply changes safely:

  1. Make small, verifiable changes
  2. Run tests after each change
  3. Commit logical units
  4. Document rationale

Step 4: Verification

Confirm success:

  • All tests passing
  • No type errors
  • Metrics documented
  • No regressions detected

Quality Gates

Pre-Refactor Checklist

Analysis Complete:
- [ ] Semantic search performed
- [ ] Relevant files read
- [ ] Complexity assessed
- [ ] Test coverage verified

Post-Refactor Checklist

Validation Complete:
- [ ] All tests passing
- [ ] No type errors
- [ ] No regressions
- [ ] Metrics improved
- [ ] Documentation updated

Mode-Specific Gates

ModeAdditional Gates
--profilePerformance improved >= 20%
--securityAll P0/P1 vulnerabilities fixed
--typesZero any, coverage >= 95%
--cleanNet code reduction achieved

Success Metrics

Universal Metrics

  • Test Status: All tests passing
  • Type Safety: No type errors
  • Complexity: Reduced from baseline
  • Coverage: Increased from baseline
  • Regressions: Zero detected
  • Documentation: Complete and accurate

Mode-Specific Metrics

ModeKey Metrics
--profileExecution time, memory usage, query count
--securityP0/P1 vulnerabilities fixed
--typesType coverage %, any count
--cleanLines removed, duplication eliminated

Mode Quick Reference

ModePurposeWhen to Use
--quickAuto-fix low-risk itemsRisk < 30, complexity < 8
--fullComplete workflow (default)Standard refactoring
--analyzeAnalysis onlyDetect and catalog issues
--reviewCreate action planPrioritize by severity
--resolveApply validated changesFrom review results
--verifyComprehensive verificationValidate results
--profilePerformance optimizationSlow functions, high memory
--securitySecurity scanOWASP compliance, audit
--typesTypeScript improvementPoor types, any usage

Scope Options

OptionPurposeExample
--scope=recentRecent changes only (default)/studio refactor --quick
--scope=file:PATHSpecific file--scope=file:src/auth/jwt.ts
--scope=directory:PATHEntire directory--scope=directory:src/features/auth
--scope=allEntire codebase--scope=all
--scope=bugBug fixing--scope=bug "TypeError in auth"

Auto-Team System

Agent teams activate automatically for complex refactors:

CriterionThresholdTeam Size
Files to analyze>= 52 agents
Modesanalyze/full/profile/security/types2-3 agents
Complexity detectedHigh2-3 agents

Disable teams: Add --no-team flag


Examples

# Quick refactor
/studio refactor --quick

# Full workflow with performance profiling
/studio refactor --full --profile

# Security audit
/studio refactor --security --scope=all

# Type safety improvement
/studio refactor --types --scope=directory:src/features

# Bug-specific refactor
/studio refactor --scope=bug "TypeError in auth"

# Team-based performance + security
/studio refactor --profile --security --team

See Also

Full documentation: @REFERENCE.md

  • Mode details (workflow, examples, success criteria)
  • Specialized modes (profile, security, types)
  • Subagent collaboration
  • Common patterns (Extract Method, etc.)
  • Configuration
  • Integration

Refactor Skill v2.1.0

What ships with it: 1 file

12.9 KB alongside SKILL.md

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.