Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create SVG graphical abstracts for journal paper submissions
.claude/skills/brycewang-stanford-graphical-abstract-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 70% | 0% |
| case-07 | ✓→✓ | = Same ✓ | 39% | 0% |
| case-08 | ✓→✓ | = Same ✓ | 32% | 0% |
A graphical abstract is a single visual summary of a research paper, designed to give readers an immediate understanding of the paper's main message at a glance. Many journals now require or strongly encourage graphical abstracts as part of the submission package, and they are increasingly important for visibility on social media and academic platforms.
This skill provides a comprehensive guide to designing effective graphical abstracts for academic papers. It covers design principles specific to scientific communication, layout strategies for common paper types, tools for creating publication-quality SVG graphics, and the specific requirements of major publishers. The emphasis is on creating clear, informative visuals that accurately represent your research without oversimplifying.
A well-designed graphical abstract can significantly increase your paper's readership and citation impact by making it more discoverable and shareable. Conversely, a poor graphical abstract—cluttered, misleading, or low-resolution—can undermine the perception of your work.
Your graphical abstract should guide the viewer's eye through the research story in a logical sequence:
Linear flow (most common):
[Input/Problem] → [Method/Approach] → [Key Finding/Result]Comparison layout:
[Before/Without] vs. [After/With]
↓ ↓
[Outcome A] [Outcome B (better)]Central concept layout:
[Aspect 1]
↑
[Aspect 4] ← [Central Concept] → [Aspect 2]
↓
[Aspect 3]Process pipeline:
[Step 1] → [Step 2] → [Step 3] → [Step 4]
↓ ↓ ↓ ↓
[Detail] [Detail] [Detail] [Result]SVG (Scalable Vector Graphics) is the ideal format for graphical abstracts because:
Draw.io (diagrams.net) — Free, browser-based, excellent for flowcharts and diagrams:
1. Open https://app.diagrams.net
2. Select "Create New Diagram"
3. Use the shape library for arrows, boxes, and connectors
4. Import your figure images as embedded elements
5. Export as SVG (File → Export as → SVG)Inkscape — Free, open-source, full vector editor:
bash# Install brew install --cask inkscape # macOS sudo apt install inkscape # Ubuntu # Set document size to journal requirements # Typical: 530 x 320 pixels (Elsevier) or custom
Figma — Collaborative design tool with a free tier:
Adobe Illustrator — Professional vector editor:
bash# Convert SVG to high-resolution PNG inkscape input.svg --export-type=png --export-dpi=300 --export-filename=output.png # Convert SVG to TIFF (for journal submission) convert -density 300 input.svg -compress lzw output.tiff # Resize to specific pixel dimensions convert output.png -resize 531x1328 -gravity center -extent 531x1328 output_elsevier.png
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 30,636 | 33,954 | +11% | 1 | 1 | 0% | 5,554 | 6,048 | +9% | 0 | 0 | — |
case-02 | fail→fail | 29,237 | 19,766 | -32% | 1 | 1 | 0% | 4,360 | 5,269 | +21% | 0 | 0 | — |
case-03 | pass→pass | 14,663 | 9,287 | -37% | 1 | 1 | 0% | 2,054 | 3,500 | +70% | 0 | 0 | — |
case-04 | fail→fail | 15,428 | 19,254 | +25% | 1 | 1 | 0% | 2,273 | 4,717 | +108% | 0 | 0 | — |
case-05 | fail→fail | 15,964 | 11,872 | -26% | 1 | 1 | 0% | 2,304 | 3,853 | +67% | 0 | 0 | — |
case-06 | fail→fail | 6,098 | 7,028 | +15% | 1 | 1 | 0% | 1,137 | 2,950 | +159% | 0 | 0 | — |
case-07 | pass→pass | 11,296 | 5,027 | -55% | 1 | 1 | 0% | 1,880 | 2,606 | +39% | 0 | 0 | — |
case-08 | pass→pass | 10,674 | 2,914 | -73% | 1 | 1 | 0% | 1,766 | 2,331 | +32% | 0 | 0 | — |
case-09 | pass→pass | 10,082 | 7,545 | -25% | 1 | 1 | 0% | 1,503 | 3,031 | +102% | 0 | 0 | — |
case-10 | pass→pass | 10,599 | 7,115 | -33% | 1 | 1 | 0% | 1,630 | 3,039 | +86% | 0 | 0 | — |
case-11 | fail→pass | 12,547 | 7,230 | -42% | 1 | 1 | 0% | 1,751 | 2,702 | +54% | 0 | 0 | — |
case-12 | pass→pass | 14,764 | 11,029 | -25% | 1 | 1 | 0% | 2,239 | 3,289 | +47% | 0 | 0 | — |
case-13 | pass→pass | 15,586 | 11,043 | -29% | 1 | 1 | 0% | 2,355 | 3,712 | +58% | 0 | 0 | — |
case-14 | pass→pass | 15,422 | 9,721 | -37% | 1 | 1 | 0% | 2,157 | 3,224 | +49% | 0 | 0 | — |
case-15 | pass→pass | 12,828 | 9,619 | -25% | 1 | 1 | 0% | 1,744 | 3,508 | +101% | 0 | 0 | — |
case-16 | pass→pass | 12,651 | 5,523 | -56% | 1 | 1 | 0% | 1,788 | 2,756 | +54% | 0 | 0 | — |
case-17 | pass→pass | 10,405 | 4,028 | -61% | 1 | 1 | 0% | 1,603 | 2,367 | +48% | 0 | 0 | — |
case-18 | pass→pass | 12,037 | 4,214 | -65% | 1 | 1 | 0% | 1,572 | 2,518 | +60% | 0 | 0 | — |
case-19 | fail→pass | 14,468 | 3,995 | -72% | 1 | 1 | 0% | 2,217 | 2,505 | +13% | 0 | 0 | — |
case-20 | pass→pass | 10,329 | 6,532 | -37% | 1 | 1 | 0% | 1,171 | 2,773 | +137% | 0 | 0 | — |
case-21 | pass→pass | 11,478 | 9,540 | -17% | 1 | 1 | 0% | 1,750 | 3,439 | +97% | 0 | 0 | — |
case-22 | pass→pass | 14,498 | 17,100 | +18% | 1 | 1 | 0% | 2,420 | 4,359 | +80% | 0 | 0 | — |
case-23 | pass→pass | 14,444 | 15,562 | +8% | 1 | 1 | 0% | 2,033 | 4,266 | +110% | 0 | 0 | — |
case-24 | pass→pass | 18,402 | 16,125 | -12% | 1 | 1 | 0% | 2,629 | 4,269 | +62% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted. The headline lift of +8 percentage points is the difference between those two pass rates over the 24 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.