AI-Ready Data

Assess data products for AI-readiness and remediate gaps. Each requirement is a self-contained directory with three markdown files per platform: check.md (context + SQL returning a 0–1 score), diagnostic.md (context + detail SQL), and fix.md (context + remediation SQL and/or organizational guidance). Each file co-locates all relevant context — constraints, gotchas, variant selection guidance, and platform-specific notes — directly above the SQL it applies to. The manifest (requirements/requirements.yaml) provides lightweight metadata for profile-load time. Every assessment has exactly six stages named after the six factors of AI-ready data — use these exact names everywhere (reports, plans, tasks): Clean, Contextual, Consumable, Current, Correlated, Compliant.

What This Skill Does

Three phases, light to deep:

1. Scan — Lightweight estate-level sweep across many schemas. Produces a comparative readiness view for prioritization. Uses the scan profile.
2. Assess — Deep evaluation of specific assets against a profile (RAG, agents, training, feature-serving, or custom). Scores each requirement, reports pass/fail.
3. Remediate — For failing requirements, present platform-specific fixes, get approval, execute, verify.

Conversation Flow