SkillProof tested Claude skills

AI Slop Cleaner

Odstraňuje AI-generovaný kódový bloat v 7 testy ověřených průchodech, ruší jakýkoli průchod, který naruší testy

od vibeeval · vibeeval/vibecosystem

Otestováno · Funguje ★ 8.4/10

AI Slop Cleaner — Odstraňuje AI-generovaný kódový bloat v 7 testy ověřených průchodech, ruší jakýkoli průchod, který naruší testy

Co umí

Post-implementation code cleanup skill that removes AI-generated bloat (dead imports, debug logs, obvious comments, dead code, one-use helpers, duplication, over-engineering) in seven risk-ordered passes, running the test suite after each pass and reverting any pass that breaks a test. Spouští se, když uživatel asks to clean up AI slop, remove code bloat, or do a hygiene pass after code generation.

Testovací report

Načteno SKILL.md (only file in the dir; pure-prompt, no scripts/deps, no security smells; frontmatter has name+description). Trigger test 5/5: "clean up the AI slop in this codebase", "just generated a feature, remove dead imports/debug logs/useless comments", "do a hygiene pass to strip AI bloat" all load it; "refactor this module to the repository pattern" (skill defers structural work to refactor-cleaner) and "format this file with prettier" correctly do not. Output test: built a 36-line slop-filled orders.ts, cleaned it one-shot as baseline vs strictly following the 7 passes plus the never-clean rules. Both stripped dead imports, console.log, commented-out cache, obvious/divider/end-block comments, the always-true `paid||paid`, and the unreachable return; the skill version additionally preserved the `// TODO: add retry logic here` comment and the try/catch boundary error handling that my baseline deleted, matching its explicit preservation rules.

Testováno: 2026-07-21 · Claude Code 2.x (agent harness)

Instalace

git clone --depth 1 https://github.com/vibeeval/vibecosystem.git /tmp/ai-slop-cleaner-src
mkdir -p ~/.claude/skills
cp -R /tmp/ai-slop-cleaner-src/skills/ai-slop-cleaner ~/.claude/skills/ai-slop-cleaner
# Pure-prompt skill: single SKILL.md, no scripts, no external deps or API keys.
# Assumes a runnable test suite in the target repo; the workflow reverts (git checkout -- .) any pass that breaks a test.
# References a kraken/spark/refactor-cleaner agent ecosystem for auto-triggering, but the manual cleanup workflow is fully self-contained.

Příkazy a ukázkové prompty

  • /ai-slop-cleanerOdstraňuje AI-generovaný kódový bloat v 7 testy ověřených průchodech, ruší jakýkoli průchod, který naruší testy

Skilly se spouštějí běžnými požadavky — žádné příkazy k zapamatování. Po instalaci ho aktivují prompty jako tyto (anglicky):

  • Clean up the bloat this agent left in the codebase
  • Run a hygiene pass after the last feature merge
  • Strip unnecessary code while keeping tests passing

Kompletní testovací report (anglicky) →