๐Ÿ† Skill Leaderboard

LLM-judged quality (1โ€“5) for each skill across Claude models โ€” scored on structure, completeness, usefulness & grounding by claude-opus-4-8.

Skillclaude-sonnet-4-6claude-haiku-4-5-20251001
competitive-analysis4.754.25
cs-health-scorecard5.004.75
executive-summary4.754.25
prd-template5.004.75
rice-prioritisation4.755.00
sprint-planning4.754.75
Average4.834.63

Higher is better (max 5). 6 skills ร— 2 models ยท generated 2026-06-18. Methodology and cases in evals/.