Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when auditing repeat-visit performance, adding offline support, or evaluating PWA readiness for a web application.
.claude/skills/thedaviddias-service-worker/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | -10% | 0% |
Service workers act as a programmable network proxy between the browser and the network. They eliminate repeated server round-trips for static assets, cut load times on repeat visits by 50–90 %, and allow the app to function at all on flaky or offline networks — which is critical for users on mobile or low-bandwidth connections.
Check whether a service worker is registered and whether it caches static assets, API responses, or provides an offline fallback.
Add a service worker registration call and implement appropriate caching strategies for static assets and navigation requests.
Explain how service workers intercept network requests and how different caching strategies trade off freshness vs. speed.
Review the service worker file and its registration. Flag missing install/ activate lifecycle handlers, absent cache-versioning, missing fetch handlers, and any patterns that could cause stale content to be served indefinitely.
For full implementation details, code examples, and framework-specific guidance, see references/rule.md.
Rule page: https://frontendchecklist.io/en/rules/performance/service-worker
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 9,358 | 6,521 | -30% | 1 | 1 | 0% | 1,766 | 1,597 | -10% | 0 | 0 | — |
case-02 | fail→pass | 10,977 | 6,559 | -40% | 1 | 1 | 0% | 1,920 | 1,591 | -17% | 0 | 0 | — |
case-03 | pass→pass | 10,217 | 7,666 | -25% | 1 | 1 | 0% | 1,965 | 1,697 | -14% | 0 | 0 | — |
case-04 | pass→pass | 11,240 | 9,106 | -19% | 1 | 1 | 0% | 2,002 | 1,975 | -1% | 0 | 0 | — |
case-05 | pass→pass | 14,016 | 9,621 | -31% | 1 | 1 | 0% | 2,486 | 2,327 | -6% | 0 | 0 | — |
case-06 | fail→pass | 15,859 | 12,880 | -19% | 1 | 1 | 0% | 2,768 | 2,620 | -5% | 0 | 0 | — |
case-07 | pass→pass | 3,210 | 2,357 | -27% | 1 | 1 | 0% | 566 | 691 | +22% | 0 | 0 | — |
case-08 | pass→pass | 8,853 | 7,732 | -13% | 1 | 1 | 0% | 1,399 | 1,633 | +17% | 0 | 0 | — |
case-09 | pass→pass | 12,623 | 11,864 | -6% | 1 | 1 | 0% | 2,287 | 2,388 | +4% | 0 | 0 | — |
case-10 | pass→pass | 15,236 | 7,066 | -54% | 1 | 1 | 0% | 2,654 | 1,650 | -38% | 0 | 0 | — |
case-11 | pass→pass | 13,376 | 7,261 | -46% | 1 | 1 | 0% | 2,637 | 1,805 | -32% | 0 | 0 | — |
case-12 | fail→pass | 12,833 | 7,959 | -38% | 1 | 1 | 0% | 2,290 | 1,867 | -18% | 0 | 0 | — |
case-13 | pass→pass | 5,654 | 3,774 | -33% | 1 | 1 | 0% | 1,049 | 1,005 | -4% | 0 | 0 | — |
case-14 | fail→fail | 15,250 | 2,244 | -85% | 1 | 1 | 0% | 2,479 | 679 | -73% | 0 | 0 | — |
case-15 | fail→pass | 11,834 | 8,450 | -29% | 1 | 1 | 0% | 2,059 | 1,852 | -10% | 0 | 0 | — |
case-16 | pass→pass | 8,563 | 3,915 | -54% | 1 | 1 | 0% | 1,548 | 1,066 | -31% | 0 | 0 | — |
case-17 | pass→pass | 3,920 | 3,443 | -12% | 1 | 1 | 0% | 714 | 960 | +34% | 0 | 0 | — |
case-18 | pass→pass | 9,026 | 7,174 | -21% | 1 | 1 | 0% | 1,700 | 1,574 | -7% | 0 | 0 | — |
case-19 | pass→pass | 13,289 | 5,794 | -56% | 1 | 1 | 0% | 2,234 | 1,301 | -42% | 0 | 0 | — |
case-20 | pass→pass | 11,336 | 11,350 | +0% | 1 | 1 | 0% | 2,197 | 2,371 | +8% | 0 | 0 | — |
case-21 | pass→pass | 10,463 | 6,732 | -36% | 1 | 1 | 0% | 1,771 | 1,479 | -16% | 0 | 0 | — |
case-22 | pass→pass | 12,258 | 9,021 | -26% | 1 | 1 | 0% | 2,168 | 1,907 | -12% | 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 +18 percentage points is the difference between those two pass rates over the 22 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.