svg-creator
Skill27 starsChanged 8 months agoApache-2.0View on GitHub
Create high-quality SVG graphics including icons, illustrations, logos, diagrams, infographics, patterns, animations, and abstract art. Use this skill whenever the user asks to create, draw, design, or generate any SVG image, vector graphic, icon, logo, badge, diagram, flowchart, infographic, pattern, or animated graphic. Also trigger when the user says "draw me", "make an SVG", "create a graphic", "design an icon", "vector illustration", or requests any visual output that would be best served as an SVG file. This skill covers both simple single-element graphics and complex multi-layered compositions with animation.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- zrosenbauer/skillstheir own version, first seen Nov 2, 2026
- Miyazaki-Kohei/claude-skillstheir own version, first seen Oct 11, 2026
- chingswy/Skill-Research-Figuretheir own version, first seen Apr 11, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- chingswy/Skill-Research-Figure★ 159same name
- zrosenbauer/skills★ 1same name
- Miyazaki-Kohei/claude-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Every SVG goes through the feedback loop script.
- Very specific: line 36
Use the view tool on /home/claude/svg_preview.png.
- Distinctive: line 41
Do proportions look right?
- Distinctive: line 91
Every non-trivial object needs: specular highlight (bright, warm), light area, half-tone (true color), form shadow (cooler hue — …
- Very specific: line 120
All gradient elements: color-interpolation="linearRGB"
- Distinctive: line 130
Add legs → render → verify connection
- Distinctive: line 131
Add arms → render → verify
- Distinctive: line 132
Add head → render → verify
- Distinctive: line 133
Add details → render → final check
- Very specific: line 135
Use thick <line> with stroke-linecap="round" for limbs.
- Very specific: line 170
Read references/advanced-techniques.md for complete recipes: filter chains, feTurbulence guide, feComponentTransfer color grading, material …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
SVG Creator Skill
THE MANDATORY WORKFLOW
Every SVG goes through the feedback loop script. No exceptions. You cannot deliver an SVG you haven't rendered and visually verified.
The Loop
WRITE SVG → RENDER → VIEW PNG → ASSESS → FIX → RENDER → VIEW → ... → DELIVERStep-by-Step
1. Write SVG to a working file:
cat > /home/claude/draft.svg << 'SVGEOF'
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 800 600">
...
</svg>
SVGEOF2. Render it (MANDATORY — the script tracks this):
python3 /path/to/skill/scripts/svg_loop.py render /home/claude/draft.svgThis creates /home/claude/svg_preview.png.
3. View the rendered PNG (MANDATORY — you must actually look):
Use the view tool on /home/claude/svg_preview.png. Study what you see.
4. Assess the result. Ask yourself:
- Are elements positioned correctly?
- Any gaps, overlaps, or misalignment?
- Do proportions look right?
- Are colors and gradients working?
- For characters: are body parts connected naturally?
5. If anything is wrong, fix the SVG and go back to step 2.
Edit /home/claude/draft.svg with str_replace or rewrite, then render and view again.
6. When it looks good, deliver:
python3 /path/to/skill/scripts/svg_loop.py finish /home/claude/draft.svg output-name.svgThis copies the SVG + preview PNG to /mnt/user-data/outputs/. The script REFUSES to deliver if you never rendered.
Then use present_files to share the output SVG with the user.
Iteration Guidelines
- Simple icons, logos, patterns: 1-2 iterations usually enough
- Diagrams, infographics: 2-3 iterations
- Scenes, illustrations: 3-5 iterations
- Characters, figures, animals: 5-8 iterations (build incrementally — torso first, then limbs one at a time, rendering after each addition)
Script Commands Reference
python3 scripts/svg_loop.py render <file.svg> # Render + view cycle
python3 scripts/svg_loop.py finish <file.svg> [name.svg] # Deliver (blocks if no render)
python3 scripts/svg_loop.py status # Check iteration count
python3 scripts/svg_loop.py reset # Start freshVISUAL QUALITY TECHNIQUES
Apply these while building the SVG. They work for all SVG types.
Multi-Stop Gradients (4+ Stops)
Two-stop gradients look flat. Use 4-8 stops with hue shifts:
<linearGradient id="sky" x1="0" y1="0" x2="0" y2="1" color-interpolation="linearRGB">
<stop offset="0%" stop-color="#0f172a"/>
<stop offset="25%" stop-color="#1e3a5f"/>
<stop offset="50%" stop-color="#3b82f6"/>
<stop offset="75%" stop-color="#93c5fd"/>
<stop offset="90%" stop-color="#fde68a"/>
<stop offset="100%" stop-color="#f97316"/>
</linearGradient>For spheres: use radial gradient with fx="0.3" fy="0.3" (offset toward light source).
Five-Zone Lighting
Every non-trivial object needs: specular highlight (bright, warm), light area, half-tone (true color), form shadow (cooler hue — blue/purple, NEVER black), reflected light (subtle warm glow on shadow edge, opacity 0.10-0.20).
Colored Shadows
Shadows use dark blue (#1a1a4e), purple (#2d1b4e), or teal (#0d3b4f). Never pure black or gray.
Drop Shadow Filter
<filter id="shadow" x="-20%" y="-20%" width="140%" height="140%"
color-interpolation-filters="linearRGB">
<feGaussianBlur in="SourceAlpha" stdDeviation="4" result="blur"/>
<feOffset in="blur" dx="3" dy="5" result="offset"/>
<feFlood flood-color="#1e1b4b" flood-opacity="0.30" result="color"/>
<feComposite in="color" in2="offset" operator="in" result="shadow"/>
<feMerge><feMergeNode in="shadow"/><feMergeNode in="SourceGraphic"/></feMerge>
</filter>Noise Texture (Breaks Digital Perfection)
<filter id="grain" color-interpolation-filters="linearRGB">
<feTurbulence type="fractalNoise" baseFrequency="0.7" numOctaves="3" stitchTiles="stitch"/>
<feColorMatrix type="saturate" values="0"/>
<feBlend in="SourceGraphic" mode="soft-light"/>
</filter>Apply at 5-15% opacity on surfaces and backgrounds.
Critical Defaults
- All
<filter>elements:color-interpolation-filters="linearRGB" - All gradient elements:
color-interpolation="linearRGB" - Animated elements:
transform-box: fill-box; transform-origin: center;
CATEGORY-SPECIFIC GUIDANCE
Characters, Figures, Animals
The hardest category. Build incrementally with aggressive feedback:
- Draw torso → render → verify
- Add legs → render → verify connection
- Add arms → render → verify
- Add head → render → verify
- Add details → render → final check
Use thick <line> with stroke-linecap="round" for limbs. Add <circle> at every joint drawn AFTER the limbs. This gives the most reliable connected look.
For animated characters: use React/JSX artifacts with forward kinematics (JS computes positions from joint angles — guarantees connected joints).
Icons (24×24 stroke-based)
Grid-aligned, 2px stroke, stroke-linecap="round", stroke-linejoin="round", fill="none", stroke="currentColor". Stay within 2-22 coordinate range. Usually one-pass works.
Logos and Badges
Geometric construction, centered. Gradient fills for depth. Max 2-3 colors. Test legibility at small size (render at 64px width).
Scenes and Landscapes
Layer back-to-front: sky → far elements (desaturated, light) → mid elements → foreground (vivid). Add atmospheric haze (semi-transparent blue-white rect) between layers. Vignette overlay last.
Data Visualizations
Compute positions from data. Use gradient fills on bars/segments. Subtle grid lines (#f1f5f9). Round caps everywhere.
Patterns
Small repeating tiles. Use <pattern> element with patternUnits="userSpaceOnUse". patternTransform="rotate(45)" for diagonal variants.
DOCUMENT STRUCTURE
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 800 600"
role="img" aria-labelledby="svg-title">
<title id="svg-title">Descriptive Title</title>
<defs><!-- gradients, filters, clipPaths --></defs>
<g id="background">...</g>
<g id="midground">...</g>
<g id="foreground">...</g>
<g id="effects">...</g>
</svg>ADVANCED REFERENCE
Read references/advanced-techniques.md for complete recipes: filter chains, feTurbulence guide, feComponentTransfer color grading, material simulation, composition templates, atmospheric effects, animation (CSS + SMIL), and more.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

