make-figures

Make-Figures Skill

You are helping a medical researcher generate publication-ready figures for medical research manuscripts. Every figure must meet journal specifications for dimensions, resolution, fonts, and color accessibility. Produce clean, data-focused visuals with no chartjunk.

Credits

The Critic Loop (Step 4b) in this skill is inspired by PaperBanana (Zhu et al., Automating Academic Illustration for AI Scientists, arXiv:2601.23265, 2025) and by prior self-refinement research — Self-Refine (Madaan et al., 2023), Reflexion (Shinn et al., 2023), and Constitutional AI (Anthropic, 2022). This is a clean-room reconstruction specialized for medical publication figures (STARD / CONSORT / PRISMA, journal-specific specs, Wong colorblind palette). No code, prompts, or configurations are derived from PaperBanana's repository.

Communication Rules

• Communicate with the user in their preferred language.
• All figure text (labels, legends, annotations) must be in English.
• Medical terminology is always in English.

Data Privacy Check

Before reading any data file, check whether it might contain Protected Health Information (PHI):

1. If *_deidentified.* files exist in the working directory, use those preferentially.
2. If only raw CSV/Excel files exist (no *_deidentified.* counterpart), warn the user (ask in the user's preferred language):

   "Does this data contain patient identifiers (names, national ID / RRN, contact details, etc.)? If so, please de-identify it first with the /deidentify skill."

3. If the user confirms the data is already de-identified or contains no PHI, proceed.

Reference Files

• Figure specifications: ${CLAUDE_SKILL_DIR}/references/figure_specs.md
• Figure style: ${CLAUDE_SKILL_DIR}/../analyze-stats/references/style/figure_style.mplstyle (or project's CLAUDE.md if available)
• Project data: See CLAUDE.md for data locations under 2_Data/

Read figure_specs.md before generating any figure to confirm journal-specific requirements.

Journal AI-Image Policies (CRITICAL — check BEFORE generation)

Synced with the user's global rule ~/.claude/rules/journal-ai-image-policies.md. The table below is the local copy used during autonomous workflow; the global rule is authoritative when conflicts arise.

Hard rule: For JACC, NEJM, or any "unknown" target journal, never use Gemini / DALL-E / Midjourney / Stable Diffusion / Nano Banana to create images that will appear in figures, Central Illustrations, or graphical abstracts. AI text-editing of the manuscript prose remains acceptable subject to standard disclosure.

Default workflow when AI images are not allowed

1. SMART Servier Medical Art — https://smart.servier.com/, CC BY 4.0, free, 3,000+ vector medical icons (anatomy, organs, ethnicity-specific human figures, drugs, devices). Commercial / journal use allowed. Required attribution (1 line in figure legend OR methods):

   Anatomical icons modified from SMART Servier Medical Art (CC BY 4.0).

2. NIAID BioArt (https://bioart.niaid.nih.gov) — public domain (US Govt), microbiology / immunology / lab-tech focus.
3. BioRender (https://www.biorender.com) — institutional license usually required; use the exported "Publication-ready" PNG/TIFF and cite per BioRender publication policy.
4. For "diseased" variants not directly available (e.g., calcified vessel from a clean vessel): reuse the healthy asset and overlay disease markers via matplotlib scatter / Circle / PathPatch. Keeps the entire pipeline non-AI and reproducible.

Asset directory convention

manuscript/figures/_assets_servier/      # CC BY 4.0 source PNGs
manuscript/figures/_assets_servier/CITATION.md   # source URL + download date per asset
manuscript/figures/_assets_data/         # data-driven raster (R / matplotlib heat maps, KM, etc.)
manuscript/figures/_legacy/              # archived prior versions

Composition scripts should load only from _assets_servier/ and _assets_data/. If a script imports from _assets_ai/, treat it as a policy violation for JACC/NEJM/unknown targets.

When a figure is produced by a data-driven .py/.R script (ROC, forest, KM, calibration, heat maps), lint that script before finalizing with the /analyze-stats code-quality gate (check_generated_code.py {script} --strict): it catches a missing plotting seed for any bootstrapped CI band, a hardcoded absolute data path, or a hand-typed data literal that should have been read from the analysis CSV.

Decoration vs information

Even when AI images are allowed, AI-generated illustrations are immediately recognizable to experienced reviewers (small decorative icons that add no information, overly uniform layouts, generic clip-art style). For high-impact submissions, prefer Servier / BioArt / BioRender + matplotlib overlays over AI.

DPI and Resolution Guide

Practical workflow for screen captures:

• Use HyperSnap or similar tool with DPI pre-set to the journal requirement
• Compose the figure in PPT at high zoom → capture at target DPI → save as TIFF/PNG
• Verify final file dimensions match journal column width requirements

Visual Abstract / Graphical Abstract

Many journals now require or strongly encourage visual abstracts. European Radiology made graphical abstracts mandatory for all Original Articles from first revision (Jan 2025). Submitting one voluntarily signals effort and can improve editorial impression.

Journal Requirements

Check the target journal profile (write-paper/references/journal_profiles/) for specific visual abstract requirements before starting.

Workflow

1. Check journal template. Look for an official PPTX template in ${CLAUDE_SKILL_DIR}/references/visual_abstract_templates/{journal}.pptx. If no journal-specific template exists, use medsci_default.pptx.
2. Extract content from the manuscript:
   • Title: Full article title
   • Hypothesis/Question: Derived from Key Point 1 or study objective (max 1 sentence)
   • Methodology: Brief flowchart or ≤3 bullets, <6 words each
   • Visual element: Study's own figure (ROC curve, flow diagram, representative image)
   • Badges: Patient cohort (N=...) | Modality/organ | Single/Multi-center
   • Main finding: Derived from Key Point 3 (<20 words)
   • Citation: Journal (year) Authors; DOI
3. Select visual element (priority order — no API needed for top options):
   1. Study's own figures (ROC, flow diagram, representative image) — always preferred
   2. Free illustration from Servier Medical Art or NIAID BioArt (see ${CLAUDE_SKILL_DIR}/references/medical_illustration_sources.md)
   3. Manual drawing in PPT/Keynote/Figma
   4. AI generation via generate_image.py --style medical (only if GEMINI_API_KEY set)
4. Generate using the script:

   python ${CLAUDE_SKILL_DIR}/scripts/generate_visual_abstract.py \
     --template medsci_default \
     --title "Article Title" \
     --hypothesis "Research question" \
     --methods "Method 1|Method 2|Method 3" \
     --finding "Main finding statement" \
     --citation "Eur Radiol (2026) Author A et al; DOI:..." \
     --visual figures/fig1_roc_curve.png \
     --badges "N=450|CT chest|Multi-center" \
     --output figures/visual_abstract.pptx
   

5. Review with user. Open the PPTX to verify layout and content. Iterate.
6. Export. PPTX is the primary deliverable. For PNG: open in PowerPoint/Keynote → export, or use LibreOffice CLI (soffice --headless --convert-to png).

Design Principles

• One page, landscape (16:9) or per journal template specification
• Three sections: Study question → Key method → Main result
• Use the study's actual figures rather than generic graphics
• Minimize text — let visuals carry the message
• Every visual element must serve a purpose (no decorative clip-art)

Available Templates

Using a journal's own template. Several journals publish one — European Radiology requires a graphical abstract from first revision and supplies EURA-GA-Jan2025.pptx. We do not redistribute them: a template you may download is not a template we may ship. Use yours directly instead:

python ${CLAUDE_SKILL_DIR}/scripts/generate_visual_abstract.py \
  --template /absolute/path/to/EURA-GA-Jan2025.pptx  ...

--template takes an absolute path to any .pptx. The script locates the fields by their text content rather than by shape name, so a journal's own template works unmodified. If the path does not exist it falls back to medsci_default.pptx.

To add a new journal template: see ${CLAUDE_SKILL_DIR}/references/visual_abstract_templates/template_guide.md.

Central Illustration vs Visual Abstract

A Central Illustration (CI) is not a Visual Abstract (VA). They serve different purposes and follow different rules. JACC family journals (JACC, JACC: Asia, JACC: Cardiovascular Imaging, JACC: Heart Failure, JACC: CardioOncology, JACC: Clinical Electrophysiology, JACC: Basic to Translational Science) require a Central Illustration with every Original Article. Reference: Fuster V, Mann DL. JACC. 2019;74(22):2816–2820.

Fuster-Mann five rules (CI must pass all)

1. Know the message. One finding, not study design + multiple findings.
2. Convey graphically, not textually. Even a simple KM curve is OK.
3. Avoid using too much text. Replace with icons or arrows.
4. Avoid secondary messages. ≤ 5 seconds for a viewer to state the main finding.
5. Simplicity is superior. Default to fewer panels.

Full guidance and validation thresholds: ${CLAUDE_SKILL_DIR}/references/jacc_central_illustration_principles.md.

CI mode invocation

python ${CLAUDE_SKILL_DIR}/scripts/generate_visual_abstract.py \
  --type central-illustration \
  --visual figures/central_illustration_v2.png \
  --citation "FirstAuthor Last et al. Journal Name 2026; vol(issue):pages." \
  --output submission/jacc_asia/central_illustration.pptx \
  --ci-zones 3 --ci-label-words 22 --ci-numerical-points 2 \
  --ci-raw-text "warranty drops to 3 years in age 45+ with cardiometabolic burden; MASLD HR 1.77"

CI mode validates before rendering and rejects (exit 2) if any of: zones > 3, label words > 30, numerical points > 4, or methodology terms (cohort flow / inclusion / exclusion / study design / enrollment / randomized / sample size / CONSORT / PRISMA / STARD) appear in --ci-raw-text. Override individual rules with --ci-allow {zones|words|numerical|methods} only when you have a defensible reason.

The JACC submission PPTX is a 10×7.5 in slide with 4 placeholders (citation textbox, content picture, footer textbox reserved, JACC logo). The red border + blue "CENTRAL ILLUSTRATION:" header are applied by JACC editorial after acceptance — authors submit only the content figure + citation.

Workflow

Step 1: Specify

Before specifying figure type, read ${CLAUDE_SKILL_DIR}/references/design_principles.md — identify (1) the one-sentence key message, (2) audience and reading-time budget, and (3) whether a figure is the right vehicle (vs a small table or in-line text). The five strategies in that file shift Step 1 from "which chart fits the data" to "what should the reader remember 10 seconds later." Skip only when the figure is mandated by a reporting guideline (e.g., PRISMA / CONSORT flow), and even then apply the cognitive-load checklist.

For reporting-guideline figures, also load ${CLAUDE_SKILL_DIR}/references/reporting_guideline_figure_map.md — the 14-row table tells you which guideline mandates which figures and whether this skill ships an official template (✅), generic flow only (⚠️), or needs manual production (❌). Critical for AI-extension guidelines (CONSORT-AI, STARD-AI, TRIPOD+AI, CLAIM 2024, DECIDE-AI).

For medical AI / engineering pipeline figures (DICOM workflow, annotation pipeline, federated learning topology, model architecture), also load ${CLAUDE_SKILL_DIR}/references/pipeline_concepts_medical_ai.md — canonical layouts, required annotations, and tool selection per type.

Optional flags:

• --study-type <type>: One of: diagnostic-accuracy, ai-validation, meta-analysis, dta-meta-analysis, observational-cohort, rct, case-report. When set, auto-generate the full figure set from the Study-Type Figure Sets table below without prompting for individual figure types.
• --data-dir <path>: Directory containing analysis outputs (CSVs, _analysis_outputs.md). Default: current working directory.

Ask the user for:

1. Figure type (from the supported types below) — skipped when --study-type is provided
2. Data source (file path, DataFrame, or manual values)
3. Target journal (for dimension/font requirements)
4. Panel layout (single panel, multi-panel, or let you decide)
5. Any special requests (annotations, highlights, reference lines)
6. Study type (if not passed via --study-type): determines the required figure set

If the user provides enough context, infer missing parameters and confirm before proceeding.

Step 2: Configure

1. Load the figure style file:

   import matplotlib.pyplot as plt
   import os
   style_path = os.path.join(os.environ.get('CLAUDE_SKILL_DIR', '.'), '../analyze-stats/references/style/figure_style.mplstyle')
   if os.path.exists(style_path):
       plt.style.use(style_path)
   

2. Look up journal-specific dimensions from ${CLAUDE_SKILL_DIR}/references/figure_specs.md.
3. Set the colorblind-safe palette (Wong palette by default).
4. Configure font sizes per element type (title, axis label, tick label, legend, annotation).

Step 3: Generate

Create the figure using Python (matplotlib/seaborn as primary, with specialized libraries as needed).

Script structure:

"""
Figure: {description}
Date: {YYYY-MM-DD}
Target: {journal}
Dimensions: {width} x {height} inches @ {DPI} DPI
"""
import numpy as np
import matplotlib.pyplot as plt
import os

style_path = os.path.join(os.environ.get('CLAUDE_SKILL_DIR', '.'), '../analyze-stats/references/style/figure_style.mplstyle')
if os.path.exists(style_path):
    plt.style.use(style_path)

# Wong colorblind-safe palette
WONG = ['#000000', '#E69F00', '#56B4E9', '#009E73',
        '#F0E442', '#0072B2', '#D55E00', '#CC79A7']

np.random.seed(42)

Step 4: Review

Present the figure to the user and ask:

• Does the layout work?
• Are labels and annotations correct?
• Any adjustments to colors, sizing, or emphasis?

Iterate until the user approves.

Step 4b: Critic Loop (self-critique before final export)

Before Step 5 Export, run the automated Critic Loop. This is two stages — deterministic quantitative checks via Python, then qualitative review by Claude itself — and the combined output tells us whether to re-render or hand off to the user.

Stage 1: Quantitative checks (critic_figure.py)

python ${CLAUDE_SKILL_DIR}/scripts/critic_figure.py \
    figures/fig1_stard.png \
    --type stard \
    --spec-min-dpi 600 \
    --spec-width-in 7.0 \
    --source-text figures/fig1_stard.txt \   # optional: expected strings for OCR coverage
    --out figures/fig1_stard.critique.json

This produces a JSON report covering:

• DPI and physical width vs. journal spec
• Dominant-color breakdown and out-of-Wong-palette fraction
• OCR-detected word count, minimum text height, and (if a source-text file was provided) source-word coverage

Stage 2: Qualitative review (Claude session)

1. Use the Read tool to load the generated PNG.
2. Read the corresponding rubric file:
   • Flow diagrams: ${CLAUDE_SKILL_DIR}/references/critic_rubrics/flow_diagram.md (sections A–G; section G adds cognitive-load and template-fidelity checks)
   • Data plots: ${CLAUDE_SKILL_DIR}/references/critic_rubrics/data_plot.md (sections A–G; section G adds calibration / fairness / colorblind+redundant / dataset-flow / decision-curve checks for medical AI papers)
   • For PRISMA / CONSORT / STARD / STROBE specifically, also read ${CLAUDE_SKILL_DIR}/references/flow_diagram_lessons.md — five production lessons covering official-template fidelity, PDF export fidelity (VML fallback), docx XML escape, sequential placeholder mapping, and frozen-version sync with the manuscript.
   • For AI-extension guidelines (CONSORT-AI, STARD-AI, TRIPOD+AI, CLAIM 2024, DECIDE-AI), also read ${CLAUDE_SKILL_DIR}/references/reporting_guideline_figure_map.md — the row for the target guideline lists mandatory figures and which ones this skill cannot template (production path documented per row).
   • For medical-AI pipeline / DICOM / federated / architecture figures, also read ${CLAUDE_SKILL_DIR}/references/pipeline_concepts_medical_ai.md.
3. Read the _why.md design notes in ${CLAUDE_SKILL_DIR}/references/exemplar_diagrams/{type}/ — hierarchy, whitespace, typography, emphasis, colour. They are the anchors. Where a rendered exemplar is bundled (template_output*.png, produced by this skill's own R script), Read 1–2 of those too; the figures cropped from published papers were removed in 2026-07 because an MIT-licensed package cannot redistribute them (see that directory's README). If you have your own exemplars locally, point the loop at them — they stay on your machine. For a non-flow data plot (forest, ROC, KM, calibration), read the matching anatomy model in ${CLAUDE_SKILL_DIR}/references/exemplar_plots/ (e.g., forest_plot.md).
4. Score every rubric item as PASS / PARTIAL / FAIL with a one-line note, using the format at the bottom of the rubric file.
5. Emit a "Required edits before next render" list of concrete source-code changes (D2 node renames, count corrections, matplotlib parameter tweaks).

Refinement loop

• If all items are PASS → proceed to Step 5 Export with critic_pass: yes.
• If any item is FAIL → apply the required edits to the source (D2 file or matplotlib script), re-render, and re-run Stage 1 + Stage 2. Default maximum is T=2 rounds; the user may request up to T=3.
• If after the max rounds some items remain PARTIAL, proceed with critic_pass: partial and record the residual items in the manifest's critic_notes field.

Record the final state in _figure_manifest.md (see the manifest format below) so downstream steps (/write-paper Phase 2 embedding and Phase 7 DOCX build) and future critic passes can see the history.

Step 5: Export

Save final outputs:

• PDF (vector format, preferred for journal submission)
• PNG (300 DPI raster, for review and presentation)
• TIFF (if the journal requires it, 300 DPI LZW compression)

Name files descriptively: fig1_roc_curve.pdf, fig2_consort_flow.pdf, etc.

For PPTX outputs (visual abstract, central illustration, or any deck the figure will live in): run the Mac-compatibility validator before delivery. PowerPoint Mac silently drops TIFF, renders <a:sp3d> 3-D bevels as red outlines that PDF export does not show, and refuses to open files whose app.xml slide count disagrees with the actual slide XML files. This script catches all four classes of defect codified in ~/.claude/rules/pptx-mac-compatibility.md:

python ${CLAUDE_SKILL_DIR}/scripts/validate_pptx_mac_compat.py \
    figures/visual_abstract.pptx \
    --json figures/visual_abstract.mac_compat.json \
    --strict

Exit code 1 means at least one FAIL — fix per the fix: field in the JSON report and re-render the PPTX before delivery. Exit code 0 with WARN is acceptable. Skip this step when the figure is PNG/PDF only (no PPTX).

Step 6: Design QC Checklist

Before delivering the final figure, verify all items:

• Font: Sans-serif (Arial/Helvetica), minimum 7pt, axis labels ≥ 9pt
• Color: Wong/Okabe-Ito colorblind-safe palette used
• Colorblind test: Would the figure work for deuteranopia? (no red-green only distinctions)
• Grayscale test: Information preserved when printed in black & white
• Alignment: All elements on a consistent grid; panels aligned
• Vector output: PDF/SVG saved (not just PNG)
• Resolution: ≥ 300 DPI for raster elements, ≥ 600 DPI for line art
• Journal specs: Dimensions, font, and format match target journal requirements
• No chartjunk: No 3D effects, unnecessary gridlines, gradient fills, or decorative elements
• Caption: Drafted with key finding, abbreviations, statistical details, and sample size

Study-Type Figure Sets

When the study type is known (from /write-paper Phase 0 or user specification), auto-detect and generate the complete required figure set without asking for each figure individually.

The manifest is mandatory. After generating all figures, write figures/_figure_manifest.md — one row per figure (Figure | Path | Type | Tool | Critic | Rounds | Description) plus a ## Critic notes section recording any residual PARTIAL items and why they were accepted. It is consumed by /write-paper Phase 2 (figure embedding) and Phase 7 (DOCX build); verify it exists and is non-empty before finishing. Format and field definitions: ${CLAUDE_SKILL_DIR}/references/figure_manifest.md.

Flow diagram generation rule. STARD / CONSORT / PRISMA / STROBE flow diagrams MUST use the standardized R pipeline scripts/generate_flow_diagram.R (DiagrammeR + Graphviz dot + rsvg) — the single canonical tool for all four. Do NOT use matplotlib FancyBboxPatch (manual coordinates break when text changes, and patches distort when embedded in DOCX). Do NOT use D2 for new flow diagrams (weak font control, overlap needs manual post-processing). Numbers in labels must be CSV-derived, or hand-written only when the value lives in a commit-tracked data artifact.

Read on demand:

Tool Selection Guide

Choose the right tool for each figure type. Using matplotlib for flow diagrams leads to hard-coded coordinates that break when text changes — use auto-layout tools instead.

Data Visualization → matplotlib/seaborn (this skill)

Best for figures where data drives the layout. This skill handles these directly:

Flow Diagrams → Dedicated Tools (NOT matplotlib)

Flow diagrams require auto-layout engines. Do NOT use matplotlib patches with manual coordinates — this causes the "absolute coordinate hell" problem where changing one box breaks all downstream positions.

R workflow for flow diagrams: See the "R flow diagram recipe" above in the Flow diagram generation rule. Key points: YAML config → Rscript scripts/generate_flow_diagram.R --type <t> --config <yaml> --out <prefix> → PDF + 300/600 dpi PNG. Templates in references/exemplar_diagrams/{strobe,consort,prisma,stard}/template_input.yaml.

Official Reporting Guideline Templates → templates/official/

When a journal requires the canonical, statement-issued template (rather than the auto-laid-out R version), use the bundled official files in templates/official/{prisma2020,consort2010,stard2015,spirit2013}/.

Refresh / fill workflow:

# Refresh from canonical sources (CC-BY 4.0 / public-statement licenses)
bash ${CLAUDE_SKILL_DIR}/scripts/fetch_official_templates.sh

# Build PRISMA 2020 .pptx (one-time; site blocks programmatic .docx fetch)
python3 ${CLAUDE_SKILL_DIR}/scripts/build_prisma2020_template.py \
    --variant new \
    --out ${CLAUDE_SKILL_DIR}/templates/official/prisma2020/PRISMA_2020_flow_new_v1.pptx

# Fill counts — positional 10-tuple matching most SR/MA workflows:
#   n_db, n_dup, n_screened, n_screen_excluded,
#   n_sought, n_assessed, n_excl_r1, n_excl_r2, n_excl_r3, n_studies
python3 ${CLAUDE_SKILL_DIR}/scripts/fill_prisma_template.py \
    --template ${CLAUDE_SKILL_DIR}/templates/official/prisma2020/PRISMA_2020_flow_new_v1.pptx \
    --counts "315,122,186,7,111,204,102,84,3,15" \
    --out fig1_prisma_filled.pptx

# Or use full JSON mapping for studies with non-standard PRISMA splits
python3 ${CLAUDE_SKILL_DIR}/scripts/fill_prisma_template.py \
    --template ${CLAUDE_SKILL_DIR}/templates/official/prisma2020/PRISMA_2020_flow_new_v1.pptx \
    --counts-file my_counts.json \
    --out fig1_prisma_filled.pptx

# STROBE — parametric single-script builder (cohort study; spine structure varies per study).
# YAML schema: stages, spine (id/stage/text), exclusions (after/text). Consecutive same-stage
# rows share one phase label automatically. Stage box fills auto-pick readable text color.
python3 ${CLAUDE_SKILL_DIR}/scripts/build_strobe_template.py \
    --config figures/figure1_strobe.yaml \
    --out    figures/figure1_strobe.pptx

The builder checks that the exclusion cascade closes — the count in a spine box, minus the exclusions declared after it, must equal the next spine box (A - Σ(exclusions after A) == B), for every link that declares an exclusion. It warns loudly on any imbalance and, with --strict-cascade, refuses to build. This catches the figure-image arithmetic drift that text-grep and prose gates miss (a dropped exclusion leaving the figure short of the analytic N). Run scripts/_strobe_cascade.py --config figure1_strobe.yaml --strict to check a config without rebuilding the diagram.

For STROBE the canonical KJR/Radiology/BMJ submission flow is:

1. Render the vector submission file via the auto-fitting Graphviz path: Rscript ${CLAUDE_SKILL_DIR}/scripts/generate_flow_diagram.R --type strobe --config figures/figure1_strobe_graphviz.yaml --out figures/figure1
2. Build the editable PowerPoint companion via build_strobe_template.py so co-authors and senior reviewers can adjust prose/positioning before sign-off.
3. Re-export the final PPTX to PDF/TIFF only after co-author edits are integrated.

See templates/official/NOTES.md for licenses, attribution, and refresh notes.

Visual / Graphical Abstracts → python-pptx Template Generator

See the Visual Abstract section above for the full workflow.

Hybrid Workflow (recommended for publication)

Data plots:    matplotlib/seaborn → PDF + PNG (this skill)
Flow diagrams: generate_flow_diagram.R (DiagrammeR + rsvg) → PDF + 300/600 dpi PNG
Final assembly: pandoc or python-docx (auto-embedded in DOCX)

Supported Figure Types (matplotlib/seaborn)

Figure Type Templates

ROC Curve

from sklearn.metrics import roc_curve, auc

fig, ax = plt.subplots(figsize=(3.5, 3.5))
fpr, tpr, _ = roc_curve(y_true, y_score)
roc_auc = auc(fpr, tpr)
ax.plot(fpr, tpr, color=WONG[5], lw=1.5,
        label=f'Model (AUC = {roc_auc:.3f})')
ax.plot([0, 1], [0, 1], 'k--', lw=0.8, alpha=0.5)
ax.set(xlabel='1 - Specificity', ylabel='Sensitivity',
       xlim=[-0.02, 1.02], ylim=[-0.02, 1.02])
ax.legend(loc='lower right', frameon=False)

• For multiple models: use distinct Wong palette colors, include AUC + 95% CI in legend.
• For comparison: report DeLong p-value in annotation.

Forest Plot

• Horizontal layout: effect sizes as squares (sized by weight), CIs as lines.
• Diamond at bottom for pooled estimate.
• Vertical dashed line at null effect (OR=1 or MD=0).
• Axis label: "Favours A | Favours B" or appropriate.
• Include heterogeneity stats (I-squared, p) below the diamond.

Flow Diagrams (STROBE / CONSORT / PRISMA / STARD)

Single canonical tool: scripts/generate_flow_diagram.R (see the R flow diagram recipe above). Do not fall back to matplotlib for flow diagrams — manual coordinates break when text changes and patches distort in DOCX. D2 remains a documented legacy fallback only when R is unavailable.

Layout invariants:

• Rectangular boxes with rounded corners for stages; notes (shape: note) for exclusion side-boxes.
• Vertical top-down flow by default; horizontal only when the manuscript layout demands it.
• Every box label contains the count (e.g., "Assessed for eligibility\n(n = 450)").
• Numbers are CSV-derived (numerical-safety) — author the YAML from an R/Python script that reads the upstream data, or cite the source file in a comment when a literal value is unavoidable.
• Follow the official template layout from each guideline.
• Use relative positioning — never hard-code absolute y-coordinates. Calculate each box position from the previous box's bottom edge plus a consistent gap constant.
• Define gap constants at the top of the script (e.g., GAP_SMALL = 1.5, GAP_BRANCH = 2.2).
• Avoid magic number padding in arrow endpoints — use named constants.

D2 approach (legacy fallback — use only when R is unavailable; the R script above is canonical):

d2 --layout elk --theme 0 flow.d2 output.svg
# Then: open SVG in Figma → grid-snap → font swap → export PDF

Caption ↔ flow-SSOT reconciliation (before Step 5 Export). The flow-diagram config (the YAML/script that generate_flow_diagram.R consumes) is the single source of truth for participant counts. A hand-written Figure 1 caption drifts from it whenever the cohort is re-locked but the caption is not — the classic "caption says n = 1,284 analytic, diagram box says n = 998" defect, which surfaces only at submission. Re-derive the caption counts from the flow config and reconcile:

python3 ${CLAUDE_SKILL_DIR}/scripts/derive_figure_legend_counts.py \
  --flow-config figures/figure1_strobe_graphviz.yaml \
  --manuscript manuscript/index.qmd \
  --out qc/figure_legend_counts.json --strict

Any n = N in the caption that is not a box count in the flow config is a MISMATCH (stale caption) — update the caption from the config, never the reverse. This pairs with numerical-safety's "re-derive prose counts every revision" rule and with /sync-submission's cross-document N checks. (The reconciler is stdlib-only and parses the config as text, so it works regardless of the flow tool.)

Calibration Plot

• 45-degree reference line (perfect calibration).
• Grouped observed vs predicted with error bars.
• Report Hosmer-Lemeshow statistic and Brier score in annotation.
• Optional: histogram of predicted probabilities at the bottom.

Kaplan-Meier Curve

• Step function with distinct colors per group.
• Censoring marks as small vertical ticks.
• Number-at-risk table below the plot (aligned with x-axis ticks).
• Log-rank p-value in annotation.
• Median survival with 95% CI if applicable.

Bland-Altman Plot

• X-axis: mean of two measurements.
• Y-axis: difference between measurements.
• Horizontal lines: mean difference (solid), +/-1.96 SD (dashed).
• Annotate the mean diff and limits of agreement values.
• Optional: proportional bias check (regression line through points).

Confusion Matrix

• Heatmap with both counts and percentages in each cell.
• Row-normalized percentages preferred (sensitivity per class).
• Clear axis labels: "Predicted" (x) and "Actual" (y).
• Use sequential colormap (Blues or Greens), not diverging.

Box/Violin Plot

• Show individual data points (jittered) overlaid on box or violin.
• Mark median and mean distinctly.
• Statistical annotation brackets with significance stars.
• Stars: * p<0.05, ** p<0.01, *** p<0.001, ns for non-significant.

Pipeline Diagram

• Horizontal or vertical flow of processing stages.
• Boxes: rounded rectangles with stage name and brief description.
• Arrows: labeled with data counts or transformation type.
• Color-code stages by category (data collection, processing, validation).
• Keep text minimal; use supplementary caption for details.

Bar Chart

• Error bars: 95% CI (preferred) or SD, stated in caption.
• Individual data points overlaid if n < 30.
• Horizontal orientation for many categories.
• Sort by value (descending) unless order is meaningful.

Heatmap

• Annotate cells with values.
• Use sequential colormap for correlation (coolwarm diverging if centered at zero).
• Mask diagonal for correlation matrices.
• Cluster rows/columns if appropriate.

Style Rules

Colors

Wong colorblind-safe palette (default):

WONG = ['#000000', '#E69F00', '#56B4E9', '#009E73',
        '#F0E442', '#0072B2', '#D55E00', '#CC79A7']

Sequential palettes (for heatmaps):

• Positive values: Blues or Greens
• Diverging (centered at 0): coolwarm or RdBu_r
• Agreement matrices: YlOrRd

Rules:

• Never use red-green only distinctions.
• Use line style (solid, dashed, dotted) in addition to color for line plots.
• Use marker shape in addition to color for scatter plots.

Typography

• Font family: Arial or Helvetica (sans-serif).
• Panel labels: uppercase bold letter, top-left of each panel.

Layout

• Minimize white space while maintaining readability.
• Align multi-panel figures on a grid.
• Consistent axis ranges across comparable panels.
• No figure titles in the plot itself (title goes in the caption below).

Statistical Annotations

• Significance stars: * p<0.05, ** p<0.01, *** p<0.001
• Place above comparison brackets.
• Report exact p-value in the figure legend or caption, not in the plot.
• For AUC, correlation, or agreement: display in the legend with 95% CI.

Journal Specifications

Default dimensions (override from figure_specs.md if journal-specific):

• Single column: 3.5 in (88 mm) width
• 1.5 column: 5.0 in (127 mm) width
• Double column: 7.0 in (178 mm) width
• Full page: 7.0 x 9.5 in (178 x 241 mm)
• DPI: 300 minimum for halftone, 600 for line art
• File formats: PDF (vector, preferred) + PNG (300 DPI)
• No chartjunk: no 3D effects, no unnecessary gridlines, no decorative elements, no gradient fills

Multi-Panel Figures

For composite figures with multiple panels:

fig, axes = plt.subplots(nrows, ncols, figsize=(width, height))

# Label each panel
for ax, label in zip(axes.flat, 'ABCDEFGH'):
    ax.text(-0.15, 1.05, label, transform=ax.transAxes,
            fontsize=12, fontweight='bold', va='top')

Common layouts:

• 2-panel horizontal: figsize=(7.0, 3.5), 1 row x 2 cols
• 2-panel vertical: figsize=(3.5, 7.0), 2 rows x 1 col
• 2x2 grid: figsize=(7.0, 7.0), 2 rows x 2 cols
• 3-panel: figsize=(7.0, 3.0), 1 row x 3 cols

Use plt.tight_layout() or fig.subplots_adjust() for spacing.

Caption Writing

After generating each figure, draft a caption following these rules:

1. First sentence: Describe what the figure shows (type + key finding).
2. Subsequent sentences: Define abbreviations, explain symbols, state sample sizes.
3. Statistical details: Note the test used and significance threshold.
4. Format: "Figure {N}. {Caption text}" -- no bold, no title case.

Example:

Figure 1. Receiver operating characteristic curves comparing the diagnostic performance of the multi-agent pipeline (blue) and single-agent baseline (orange) for identifying incorrect Anki flashcard content. The area under the curve was 0.92 (95% CI: 0.89-0.95) for the multi-agent pipeline and 0.84 (95% CI: 0.80-0.88) for the single-agent baseline (DeLong test, p = 0.003). The dashed diagonal line represents chance performance.

Skill Interactions

Error Handling

• If data is insufficient for the requested figure type, explain what is needed and ask the user.
• If a figure exceeds journal dimension limits, resize and report the adjustment.
• If text overlaps in the figure, try tight_layout(), reduce font size, or adjust spacing.
• Never fabricate data points. If sample data is needed for a template demo, explicitly label it as "example data."

CLI Tools Available

ImageMagick, Ghostscript, FFmpeg are installed and can be used for post-processing:

# Figure DPI/format conversion for journal submission
magick input.png -density 300 -units PixelsPerInch output.tiff
magick input.png -resize 1200x -quality 95 output.jpg

# CMYK conversion (some print journals require this)
magick input.png -colorspace CMYK output.tiff

Portal-ready TIFF (SNAPP .png-not-accepted / 25 MB cap)

A raw magick ... output.tiff keeps the alpha channel (transparent regions print black on many production pipelines) and stays uncompressed (a 600-dpi RGBA TIFF blows past a portal's 25 MB cap). export_portal_tiff.py does the flatten-and-compress a human otherwise does by hand and verifies the result is pixel-identical to that white-flatten before handing it over — use it when a portal accepts only .tiff/.jpeg/.eps (Springer Nature SNAPP) or caps figure size (JACC: Asia):

python3 scripts/export_portal_tiff.py --in figure.png --out figure.tiff --max-mb 25
# LZW-compressed, RGBA→RGB white-flattened, pixel-identity-verified; exit 1 if still over the cap

# Multi-panel figure assembly (A/B/C/D panels)
magick montage panelA.png panelB.png panelC.png panelD.png \
  -tile 2x2 -geometry +10+10 -density 300 combined.png

# Animated figure (GIF from frame sequence)
ffmpeg -framerate 2 -i frame_%03d.png -vf "scale=800:-1" output.gif

# Video from figure sequence (for supplementary materials)
ffmpeg -framerate 1 -i slide_%03d.png -c:v libx264 -pix_fmt yuv420p supplementary_video.mp4

AI Image Generation (Optional)

AI illustration is a supplementary option, not a requirement. Visual abstracts and figures can be completed without any API key using study figures and free illustration libraries.

If GEMINI_API_KEY is set, the generate_image.py script can generate illustrations:

python ${CLAUDE_SKILL_DIR}/scripts/generate_image.py \
  "Clean medical illustration of a CT-guided lung biopsy procedure, \
   flat vector style, white background, no text" \
  --output output.png --aspect 16:9

Use for: procedural schematics, anatomical illustrations, pipeline diagrams. Always review AI output against the AI-Generated Figure Warning section above.

If GEMINI_API_KEY is not set, guide the user to free illustration resources: see ${CLAUDE_SKILL_DIR}/references/medical_illustration_sources.md.

Language

• Code and figure text: English
• Communication with user: Match user's preferred language
• Medical terms: English only

Anti-Hallucination

• Never fabricate references. All citations must be verified via /search-lit with confirmed DOI or PMID. Mark unverified references as [UNVERIFIED - NEEDS MANUAL CHECK].
• Never invent clinical definitions, diagnostic criteria, or guideline recommendations. If uncertain, flag with [VERIFY] and ask the user.
• Never fabricate numerical results — compliance percentages, scores, effect sizes, or sample sizes must come from actual data or analysis output.
• If a reporting guideline item, journal policy, or clinical standard is uncertain, state the uncertainty rather than guessing.

Global-rule references

Some passages in this skill cite a path of the form ~/.claude/rules/<name>.md. Those are the maintainer's personal global rules, kept outside this repository. They are not shipped with this skill and will not exist on your machine; they appear only as provenance for where a convention came from. If one of them looks like it is standing in for an instruction you actually need, that is a bug — please open an issue, because the instruction belongs here.