SkillProof tested Claude skills

MECE

Scores any spec/plan/PRD for overlaps and gaps (Mutually Exclusive, Collectively Exhaustive) and asks before fixing.

Autor: uxderrick · uxderrick/mece-skill

Testowano · Działa ★ 8.8/10

MECE — Scores any spec/plan/PRD for overlaps and gaps (Mutually Exclusive, Collectively Exhaustive) and asks before fixing.

Co robi ten skill

Validates an existing spec, plan, PRD, API design, or category list against the MECE framework -- every item in exactly one bucket, no gaps -- producing a 0-100 score with severity-tagged overlap/gap findings, or decomposes a fresh topic into a MECE issue tree. Triggers on 'check this', 'anything missing?', 'is this complete?', or 'break this down'; always reports and asks before applying any fix.

Raport z testu

Ran a fresh 6-item notification-system spec (not one of the skill's own worked examples) through both a baseline and a skill-guided pass: the baseline gave a vague 'looks pretty solid, maybe add SMS' reaction and never caught that 'Notification preferences' and 'User notification settings' were the same concept under two names; the skill's Step 2 mutual-exclusivity test caught that overlap explicitly, scored the spec 50/100, tagged two critical gaps (no SMS/text channel, no unsubscribe/compliance layer) and two moderate gaps (batching/throttling, template management), then correctly stopped and asked before fixing per its own hard rule. Fully self-contained -- SKILL.md, framework.md, and examples.md all live inside skills/mece/ with zero external or plugin-host dependencies, so the plain git-clone-and-cp-r install actually works as documented.

Testowano: 2026-07-15 · Claude Code 2.x (agent harness)

Instalacja

git clone https://github.com/uxderrick/mece-skill.git
mkdir -p ~/.claude/skills
cp -r mece-skill/skills/mece ~/.claude/skills/mece

Komendy i przykładowe prompty

  • /meceScores any spec/plan/PRD for overlaps and gaps (Mutually Exclusive, Collectively Exhaustive) and asks before fixing.

Skille uruchamiają się na zwykłe polecenia — bez komend do zapamiętania. Po instalacji aktywują go prompty takie jak te (po angielsku):

  • Can you check if this PRD has any gaps or overlapping requirements in it?
  • Is my API design missing anything important before I send it for review?
  • Break this messy feature list down into clean, non-overlapping categories

Pełny raport →