Conciseness

The body is lean: terse default tables, one-line format conventions with non-obvious justifications ("prefer CRAM over BAM (30-50% smaller)"), and no explanations of concepts Claude already knows. Every line is a decision, a convention, or a pointer.

Actionability

Concrete tool choices for every scenario (BWA-MEM2, STAR, Minimap2, DeepVariant, Mutect2, Manta, GRIDSS, VEP), concrete commands ("always use bgzip + tabix for indexing"), and a copy-paste-ready `nextflow run nf-core/sarek` example with explicit escape hatches for GATK-ecosystem and somatic cases.

Workflow Clarity

Routing is clear (defaults → file formats → workflow-type dispatch to references → nf-core table) and validation guidance exists ("Validate files before downstream analysis", "QC at every stage", "set clear PASS/FAIL thresholds"), but the end-to-end sequence is implied by section order rather than enumerated with explicit validate-and-proceed checkpoints. Not 3: checkpoints are stated, not absent; not 5: no explicit feedback loop for failed validation.

Progressive Disclosure

The body is a genuine overview: detailed per-workflow material is split into three well-signaled, one-level-deep references ([references/rnaseq.md], [references/annotation.md], [references/cnv.md]), all of which exist in the bundle, while defaults, conventions, and the nf-core table stay appropriately inline.

Specificity

The description lists multiple concrete capabilities — "Provides opinionated tool defaults (aligners, variant callers, annotators), genomics file-format and reproducibility conventions, an nf-core pipeline selection table, and detailed references for RNA-seq (bulk and single-cell), variant annotation with clinical interpretation, and CNV/SV analysis" — with comprehensive coverage and no vague filler.

Completeness

It explicitly answers both questions: "This skill should be used when the user asks to..." gives concrete when-triggers, and "Provides opinionated tool defaults... conventions... selection table, and detailed references..." gives a clear what, matching the top anchor's structure.

Trigger Term Quality

Nine quoted natural trigger phrases ("build a genomics pipeline", "pick an nf-core pipeline", "which aligner/variant caller should I use", "analyze single-cell RNA-seq", "interpret variants with ACMG criteria", "detect CNVs", "call structural variants") plus synonyms (NGS, CNV/SV) give excellent coverage, but no file-extension triggers (.fastq, .bam, .vcf) that users commonly mention, placing it between anchors 4 and 5.

Distinctiveness Conflict Risk

A clear genomics-pipeline niche with distinct triggers and an explicit non-goal boundary — "It does not teach Nextflow DSL2, Snakemake, or WDL authoring — for writing workflow code by hand, use those frameworks' own documentation" — redirects overlapping requests elsewhere, minimizing conflict risk. Voice is consistently third person.