agentsclimarketplace

Simplify

Skill ngocsangyem/MeowKit/.claude/skills/simplify

Production ready. AI Agent Workflow System for Claude Code

Install
npx -y skills add ngocsangyem/MeowKit --skill simplify

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

  • 15 stars15 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.

What its author says it does

Copied from the file, not written here

Post-implementation simplification pass. Reduces complexity without changing behavior. Use after Phase 3 (Build GREEN) and before Phase 4 (Review). Catches dead code, unnecessary abstractions, over-engineering, and redundant patterns. Triggers on "simplify", "clean up", "reduce complexity", "too complex". NOT for style/convention enforcement (see mk:clean-code); NOT for reviewing diffs or PRs (see mk:review).

SKILL.md

3.3 KB, as published. Nobody here has run it

Code Simplification

Runs after implementation, before review. Reduces complexity while preserving behavior.

For ad-hoc quality review outside the current implementation diff, use mk:clean-code. mk:simplify is behavior-preserving and scoped to the current diff only.

Iron Rule

Behavior must not change. Every simplification must pass the exact same tests as before. If tests fail after simplification, the simplification was wrong — revert it.

What to Look For

Remove

  • Dead code — functions never called, variables never read, unreachable branches, import never used
  • Commented-out code — if it's in git history, delete it from source
  • Unnecessary abstractions — wrapper that adds no value, interface with one implementation
  • Redundant null checks — checking null after a guard that already prevents null

Simplify

  • Deep nesting — 3+ levels of if/else → extract early returns or guard clauses
  • God functions — >50 lines → extract focused helpers
  • Duplicate logic — same pattern in 3+ places → extract shared utility
  • Complex conditionsif (a && !b || (c && d)) → extract to named boolean or function

Don't Touch

  • Working code that's "not how I'd write it" — style is not complexity
  • Performance-optimized code — it looks complex for a reason
  • Code outside the current diff — scope discipline, even for simplification

Process

  1. Identify — scan changed files for complexity signals
  2. Propose — list each simplification with before/after preview
  3. Apply — make changes one at a time
  4. Verify — run full test suite after each change (not batched)
  5. Report — list what was simplified and why

Gotchas

  • "Just one more cleanup" — scope creep is the #1 risk. Only simplify code from the current diff
  • Removing "dead" code that's used via reflection/dynamic import — grep for string references, not just static imports
  • Extracting too early — 2 copies is not enough to justify a shared utility. Wait for 3
  • Breaking public API — simplifying internals is fine; changing exported signatures is a breaking change
  • Test-only code — don't simplify test helpers/fixtures; they prioritize readability over DRY

Pipeline Position

Phase 2 (Test — RED if `--tdd`) → Phase 3 (Build) → [mk:simplify] → Phase 4 (Review)

Simplification happens AFTER tests pass but BEFORE review. Reviewer sees the simplified code.

Keep looking

Skills are one crate of 328,083. 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.