skill-repository-curation Case #4

# Skill Repository Consolidation The directory `/workspace/environment/skills/sales-data-pipeline` is a fragmented knowledge base: **14 sub-skills** with substantial overlap and redundancy. Use the `skill-creator` skill to clean it up. ## Your Goal Consolidate the 14 sub-skills into a smaller set of **4–6 cohesive sub-skills** that together cover the full data-pipeline lifecycle — ingestion, cleaning, validation, transformation, aggregation/reporting, and export. Eliminate overlapping or redundant skills; preserve every distinct capability of the originals. ## Deliverables 1. **Consolidated skill directory** at `/workspace/environment/skills/sales-data-pipeline/` — replace the 14 original sub-skills with 4–6 well-named sub-skills, each carrying its own well-formed `SKILL.md`. 2. **Consolidation rationale** at `/workspace/environment/skills/sales-data-pipeline/CONSOLIDATION_RATIONALE.md` — a written record of: - Which original sub-skills were merged, and into what. - Why each merge made sense (overlap, redundancy, subset/superset, etc.). - How coverage of the original capabilities is preserved. This file is **scored**: the verifier uses it to determine whether you identified redundancy pairs and articulated a clear merge rationale. ## Scoring Dimensions (each contributes to total / 100) | Dimension | Weight | Source | |---|---|---| | Redundancy identified | 15 | `CONSOLIDATION_RATIONALE.md` + your reasoning trace | | Coverage preserved | 20 | union of consolidated `SKILL.md` files | | Skill count reduced | 15 | count of sub-skills (target 4–6) | | Overlap eliminated | 15 | Jaccard similarity across consolidated skills | | Pipeline coherent | 15 | stage coverage (ingestion, cleaning, validation, transformation, aggregation, export) | | Skill quality | 10 | each `SKILL.md` well-formed | | Consolidation rationale | 10 | `CONSOLIDATION_RATIONALE.md` + your reasoning trace | Aim for ≥80/100.