AravindS-Wick Lite Superpowers
Token-optimized code intelligence. Replaces superpowers@claude-plugins-official for code tasks — 75–85% fewer tokens, no forced Opus 4.6.
Quick Start
/lite-superpowers --analyze # Analyze current file/selection
/lite-superpowers --refactor # Suggest refactoring improvements
/lite-superpowers --optimize # Performance & efficiency improvements
/lite-superpowers --generate # Boilerplate / scaffold code
/lite-superpowers --explain # Explain complex code sections
/lite-superpowers --review # Implementation review with rationale
Features
| Flag | What it does | Tokens |
|---|---|---|
--analyze |
Code quality: complexity, coupling, smells | 200–400 |
--refactor |
Before/after with clear rationale | 300–500 |
--optimize |
O(n) complexity, memory, bundle size | 200–400 |
--generate |
Scaffold from spec or pattern | 300–600 |
--explain |
Plain-language breakdown of logic | 150–300 |
--review |
Senior-level implementation critique | 300–500 |
Scope Flags
/lite-superpowers --analyze --quick # Current file only
/lite-superpowers --analyze --full # Whole codebase
/lite-superpowers --analyze --file=src/auth.js
What It Catches
Code quality: God objects, spaghetti coupling, duplicated logic, dead code, inconsistent naming
Performance: O(n²) loops, unnecessary re-renders, memory leaks, large bundle imports, N+1 queries
Security surface: Missing input validation, unescaped output, hardcoded values, missing error boundaries
Patterns: Suggests Factory, Observer, Strategy, Repository where applicable — shows before/after
Token Comparison
| Task | Full superpowers | Lite superpowers | Savings |
|---|---|---|---|
| Single file analyze | ~2,000 | ~350 | 83% |
| Refactor suggestion | ~3,000 | ~450 | 85% |
| Code review | ~4,000 | ~500 | 88% |
| Boilerplate generate | ~2,500 | ~500 | 80% |
Integration
/lite-superpowers --analyze → /dev-assistant --feature=architecture (structural issues)
/lite-superpowers --review → /pr-check --checks=linting (before PR)
/lite-superpowers --generate → /test-global --checks=jest (test generated code)
Token avg: 200–600 per command | Model: Sonnet (never forces Opus)