Mostly efficient and packed with valuable non-obvious param gotchas (e.g. 'param is variant_id, not rsid', 'assay_title="TF ChIP-seq"'), but the histone-mark primer ('H3K27ac signals active enhancers...') restates biology Claude already knows and could be trimmed.

Gives exact tool names, exact parameter names with common-mistake callouts, per-phase reasoning tips, fallback strategies, and concrete example workflow calls — fully executable, copy-paste-ready guidance.

A clear Phase 0–6 sequence with a workflow diagram, per-phase reasoning checkpoints, an explicit evidence-grading synthesis step, and fallback error-recovery loops for empty results.

The body is well-sectioned and navigable, but it is a monolithic single file with no bundle references; the detailed phase docs, example workflows, and fallback strategies are inline content that could be split into one-level-deep reference files.

Lists multiple concrete actions — 'GWAS association lookup, eQTL evidence (GTEx), chromatin state (ENCODE), regulatory variant scoring (RegulomeDB, CADD), and TF-binding disruption' — matching the score-3 anchor of several specific actions.

Explicitly answers both what (the five analysis actions) and when via the 'Use for non-coding GWAS hit interpretation, eQTL-based gene assignment...' trigger clause, satisfying the score-3 anchor.

Natural domain terms a genomics user would actually say appear throughout — 'non-coding GWAS hit interpretation', 'eQTL-based gene assignment', 'regulatory mechanism reasoning' — giving good coverage rather than generic jargon.

The non-coding/regulatory niche is sharply scoped and 'Distinct from coding-variant tools' explicitly separates it from sibling variant skills, making wrong-skill triggering unlikely.