figure-style

Publication-grade figure correctness and legibility rules. Load before drawing any plot and call `apply_figure_style()` — sets a role-mapped font-size ladder, outward ticks, frameless legends, and 300-dpi output. The skill is a checklist, not a house look: data fidelity (claim-titles tested against every row, excluded data never enters summaries), label economy (floor and ceiling), colour threading, chart-choice-by-data-shape, layout, and a render-then-verify QA loop (bbox collision + per-panel perceptual check). Ships helpers: focal_palette, bar_with_points, strip_with_median, end_of_line_labels, panel_letter, set_frame, panel_crops. For multi-panel figures load `figure-composer`; for whole-paper figure arc load `paper-narrative`.

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A tightly-written, highly actionable ruleset with a strong render-then-verify feedback loop and real executable code. Its weaknesses are the absence of any bundle files to split out the helper implementations, and minor verbosity/over-restatement in a few rules.

Suggestions

Move the helper implementations (focal_palette, bar_with_points, strip_with_median, end_of_line_labels, panel_letter, set_frame, panel_crops) into a referenced scripts/ file and link to it, so SKILL.md stays a lean overview and progressive disclosure goes from 4 to 5.

Trim restated rules (e.g. comparator-naming appears in both §2.1 and §5.6, and the closing maxim re-states §2 and §4) to tighten conciseness.

Add the concrete matplotlib call alongside rules that currently only name a check (e.g. the CVD check in §4.5 and the contrast check in §5.5) so every rule has an executable action.

Dimension	Reasoning	Score
Conciseness	The body is dense and rule-packed with almost no padded concept explanation (no 'what is a PDF'-style filler), but it is long and a few rule comments and the closing maxim restate ideas, leaving minor trims available — efficient but not maximally lean.	4 / 5
Actionability	Provides concrete, executable artifacts: real matplotlib calls (ax.margins(0.04), apply_figure_style(sizes=(8,7,6)), panel_letter(ax, 'a', case=...)) and a complete bbox-collision check snippet; minor gaps are that several rules describe what to check without the exact call to perform it.	4 / 5
Workflow Clarity	Sequences the work explicitly (§0 scope, §1–§9 rules, then §9 render-then-verify) and §9 gives an ordered feedback loop: savefig -> geometric bbox check (assert overlaps==[]) -> fix and re-save until clean -> per-panel perceptual crop check, with explicit 'only finalize after clean' gating.	5 / 5
Progressive Disclosure	Well-organized into nine numbered sections with a clear scope header and forward references to figure-composer and paper-narrative, but the document is a single monolithic SKILL.md with no references/ scripts/ or assets/ bundle files, so the helper inventory (focal_palette, bar_with_points, etc.) lives only in prose rather than being split into referenced files.	4 / 5
	Total	17 / 20 Passed

Description

80%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A dense, specific description that names concrete helpers, rules, and explicit load triggers, and cleanly separates this skill's tier from figure-composer and paper-narrative. Its main gap is the absence of a clean 'Use when...' clause framed as user-facing scenarios, which keeps completeness at 4 rather than 5.

Suggestions

Add an explicit 'Use when drawing or revising publication plots, multi-panel figures, or scientific charts for a paper' clause so the when/trigger is framed as user-request scenarios rather than only a load instruction.

Soften internal shorthand (e.g. 'colour threading', 'claim-titles') or pair it with a plain-language synonym so a user's natural phrasing maps onto the description more reliably.

Lead with the one-line trigger before the helper inventory so the activation signal is front-loaded.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions and named helpers (apply_figure_style, focal_palette, bar_with_points, end_of_line_labels, panel_letter, panel_crops) plus specific rule families (role-mapped font-size ladder, outward ticks, frameless legends, 300-dpi, bbox collision + per-panel perceptual check) — comprehensive and concrete coverage.	5 / 5
Completeness	Clearly answers 'what' (correctness/legibility checklist with named rules and helpers) and gives explicit load-trigger guidance ('Load before drawing any plot'), but the 'when' is framed as a load instruction rather than a full 'Use when...' clause naming user-request scenarios, so it is present and specific but could be more explicit.	4 / 5
Trigger Term Quality	Includes natural trigger phrasing like 'Load before drawing any plot', 'figure', 'multi-panel figures', and 'whole-paper figure arc', but relies on domain shorthand (e.g. claim-titles, colour threading) rather than the most common lay synonyms a user would say, so a few natural terms are missing.	4 / 5
Distinctiveness Conflict Risk	Niche is well-scoped to publication-grade figure correctness/legibility and explicitly carves out figure-composer (multi-panel) and paper-narrative (whole-paper arc) as separate skills, giving it mostly-distinct triggers with only minor overlap risk against those related skills.	4 / 5
	Total	17 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: UnicomAI/wanwu
Path: configs/microservice/bff-service/configs/agent-skills/claude-science/figure-style/SKILL.md
Commit: 030c20b

Reviewed: 15 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.