Conciseness

The body is dense, imperative, and assumes competence — no basic-concept padding, no library tutorials — but not every token earns its place: the §8 anti-pattern list largely restates rules already stated in §1–§7 ("Filled bars on a log-scaled value axis" repeats §3.3; the diverging-colormap center bullet repeats §4.4), and the §0 scope discussion overlaps the intro paragraph. That is the efficient-with-minor-trimming anchor, not the lean 5.

Actionability

Two complete, executable code blocks (the §9.1 bbox/overlap check and the §9.2 panel-crop loop) plus named helper calls with concrete signatures ("apply_figure_style(sizes=(8,7,6))", "focal_palette(...)", "end_of_line_labels(...)", "panel_crops(fig)"). Rules are phrased as testable criteria ("read it aloud cold — if the listener asks 'fewer genes what?', rewrite"), which is fully actionable guidance rather than vague direction.

Workflow Clarity

The process is explicitly sequenced with checkpoints: scope gate in §0, the apply-style call before plotting, per-section rules while plotting, then "After `fig.savefig(...)`, before `save_artifacts`" for §9, with a genuine feedback loop — "Fix (move, shorten, stagger) and re-save until clean" — and a second perceptual verification pass. This matches the clear-sequence-with-explicit-validation-and-feedback anchor.

Progressive Disclosure

Section structure (§0–§9) is clear and easy to navigate, but all 409 lines live inline in SKILL.md with no bundle files at all — the §6 chart-family catalog and the §9 QA code are substantial bodies that would plausibly belong in one-level-deep reference files. This is the some-structure-but-content-that-should-be-separate-is-inline anchor; it is above 2 because the in-file organization is strong, not minimal.

Specificity

"Covers data fidelity, label economy, color threading, chart choice, layout, and render-then-verify QA" names the domain plus six concrete coverage areas, matching the several-specific-actions anchor. It stays below the top anchor because capabilities are listed as abstract nouns rather than the concrete actions a 5 exemplifies.

Completeness

It explicitly answers both questions: what ("Publication-grade correctness and legibility rules for final-deliverable scientific figures... Covers data fidelity, label economy, color threading, chart choice, layout, and render-then-verify QA") and when ("Use for a figure that will ship in a report, paper, export, or kept artifact"), plus a negative boundary ("not exploratory plots"). This matches the anchor for clearly and explicitly answering both with concrete trigger phrases.

Trigger Term Quality

Natural trigger terms are present — "figure," "report," "paper," "export," "publication" — and the phrase "Use for a figure that will ship in a report, paper, export, or kept artifact" reads like what a user would actually say. A few common synonyms (e.g., "journal," "manuscript," "publication-ready") are missing, keeping it at 4 rather than 5.

Distinctiveness Conflict Risk

The description carves out a clear niche (final-deliverable scientific figures) and explicitly hands off adjacent scopes — "For multi-panel composition use `figure-composer`; for whole-paper ordering use `paper-narrative`" — while excluding exploratory plots. This is exactly the clear-niche, minimal-conflict top anchor.