Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build interactive web artifacts — single-file HTML/CSS/JS apps, data visualizations, dashboards, games, and interactive demos. Use when asked to create something visual and interactive that runs in a browser.
.claude/skills/kxiandaoyan-web-artifacts-builder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-20 | ✓→✗ | ▼ Worse | 79% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -7% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | 58% | 0% |
Create self-contained, interactive web artifacts delivered as single HTML files.
Always produce a single index.html file saved to /data/ that includes all CSS and JS inline. The file must work without any server or build step.
html<!DOCTYPE html> <html lang="en"> <head> <meta charset="UTF-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <title>App Title</title> <style> /* All CSS inline */ </style> </head> <body> <!-- Content --> <script> // All JS inline </script> </body> </html>
Use Chart.js (CDN) for charts. Load data inline as JS arrays.
html<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
Tailwind CSS (CDN) for rapid layout:
html<script src="https://cdn.tailwindcss.com"></script>
Vanilla JS + Canvas API. No dependencies needed.
HTML form elements + JS event listeners. Validate client-side.
/data/artifact-name.html or /data/index.html/data/ using the write tool| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | 28,122 | 52,059 | +85% | 1 | 1 | 0% | 6,187 | 7,697 | +24% | 0 | 0 | — |
case-10 | fail→pass | 26,901 | 29,679 | +10% | 1 | 1 | 0% | 6,157 | 6,612 | +7% | 0 | 0 | — |
case-11 | fail→fail | 28,997 | 27,788 | -4% | 1 | 1 | 0% | 6,165 | 6,620 | +7% | 0 | 0 | — |
case-12 | fail→fail | 20,716 | 53,183 | +157% | 1 | 1 | 0% | 4,599 | 8,243 | +79% | 0 | 0 | — |
case-02 | fail→fail | 26,738 | 8,396 | -69% | 1 | 1 | 0% | 6,181 | 1,639 | -73% | 0 | 0 | — |
case-01 | fail→fail | 27,849 | 11,618 | -58% | 1 | 1 | 0% | 6,183 | 2,454 | -60% | 0 | 0 | — |
case-04 | fail→fail | 22,576 | 53,144 | +135% | 1 | 1 | 0% | 5,064 | 7,475 | +48% | 0 | 0 | — |
case-05 | fail→pass | 27,801 | 27,495 | -1% | 1 | 1 | 0% | 6,164 | 6,619 | +7% | 0 | 0 | — |
case-06 | fail→fail | 27,756 | 53,516 | +93% | 1 | 1 | 0% | 6,172 | 8,637 | +40% | 0 | 0 | — |
case-07 | fail→fail | 24,667 | 57,880 | +135% | 1 | 1 | 0% | 5,410 | 9,278 | +71% | 0 | 0 | — |
case-08 | fail→fail | 31,000 | 55,430 | +79% | 1 | 1 | 0% | 6,162 | 8,571 | +39% | 0 | 0 | — |
case-09 | fail→fail | 27,951 | 54,166 | +94% | 1 | 1 | 0% | 6,161 | 8,095 | +31% | 0 | 0 | — |
case-13 | fail→fail | 28,775 | 29,582 | +3% | 1 | 1 | 0% | 6,132 | 6,620 | +8% | 0 | 0 | — |
case-14 | fail→fail | 27,674 | 29,657 | +7% | 1 | 1 | 0% | 6,161 | 6,616 | +7% | 0 | 0 | — |
case-15 | fail→fail | 27,388 | 56,372 | +106% | 1 | 1 | 0% | 6,161 | 10,633 | +73% | 0 | 0 | — |
case-16 | fail→fail | 22,356 | 56,231 | +152% | 1 | 1 | 0% | 4,918 | 10,136 | +106% | 0 | 0 | — |
case-17 | fail→fail | 28,087 | 66,465 | +137% | 1 | 1 | 0% | 6,159 | 2,775 | -55% | 0 | 0 | — |
case-18 | fail→fail | 28,053 | 54,865 | +96% | 1 | 1 | 0% | 6,127 | 8,725 | +42% | 0 | 0 | — |
case-19 | fail→fail | 21,520 | 21,696 | +1% | 1 | 1 | 0% | 4,740 | 4,406 | -7% | 0 | 0 | — |
case-20 | pass→fail | 11,850 | 21,459 | +81% | 1 | 1 | 0% | 2,364 | 4,235 | +79% | 0 | 0 | — |
case-21 | pass→pass | 19,284 | 29,413 | +53% | 1 | 1 | 0% | 3,748 | 5,921 | +58% | 0 | 0 | — |
case-22 | pass→fail | 14,144 | 11,892 | -16% | 1 | 1 | 0% | 2,700 | 2,523 | -7% | 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. 22 cases were attempted. The headline lift of -29 percentage points is the difference between those two pass rates over the 22 comparable cases. 9 cases got worse with the skill loaded, and they are included in that figure.
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.