Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build promo videos for local businesses in HyperFrames -- the new-menu drop, the seasonal offer, the event announcement, the "we're hiring" post, the Google Business Profile clip. Restaurant, salon, gym, trades: 15-30 seconds, brand-clean, captioned, sized for Instagram, Facebook, and the profile.
.claude/skills/onewave-ai-hyperframes-local-promo/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -1% | 0% |
Local businesses need a steady drip of short promo video and have zero production budget -- this skill is the in-house video guy. It directs quick, renderable HyperFrames promos from whatever the business has: phone photos, the menu PDF, the offer details, the logo. Composition mechanics live in the hyperframes skill (hyperframes-cli for init/lint/preview/render, hyperframes-media for voiceover and background removal on product shots).
Promo type -- offer/sale, new item or service, event, announcement (new hours, new location), hiring, or evergreen brand clip -- plus the one detail that matters (the offer terms, the date, the dish), the assets on hand (photos, logo, menu, brand colors), and where it runs. Local promos are volume content: get to the build in minutes, not meetings.
The local promo formula, 15-30 seconds:
Photos treated with respect: real phone photos beat stock every time for local trust -- crop and pan them with intent (hyperframes-media background removal to lift product shots onto brand color when it helps). No stock footage of generic smiling people; skip stock entirely rather than fake it (stock-photo-finder only for texture/ambience gaps).
The second promo should take a tenth of the setup: save the brand tokens, logo lockup scene, and close-scene as the business's template, and offer the content calendar angle -- one promo per week (offer, item spotlight, review feature via hyperframes-testimonial-builder, behind-the-scenes) beats one perfect video per quarter.
job-post-writer for the copy)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 17,791 | 20,572 | +16% | 1 | 1 | 0% | 2,946 | 4,506 | +53% | 0 | 0 | — |
case-02 | fail→fail | 17,934 | 18,968 | +6% | 1 | 1 | 0% | 3,796 | 4,895 | +29% | 0 | 0 | — |
case-03 | fail→fail | 19,726 | 27,467 | +39% | 1 | 1 | 0% | 3,375 | 5,480 | +62% | 0 | 0 | — |
case-04 | fail→pass | 12,604 | 14,591 | +16% | 1 | 1 | 0% | 2,178 | 3,354 | +54% | 0 | 0 | — |
case-05 | fail→fail | 10,263 | 11,716 | +14% | 1 | 1 | 0% | 1,638 | 2,787 | +70% | 0 | 0 | — |
case-06 | fail→pass | 10,715 | 8,482 | -21% | 1 | 1 | 0% | 1,745 | 2,212 | +27% | 0 | 0 | — |
case-11 | pass→pass | 17,486 | 15,325 | -12% | 1 | 1 | 0% | 2,839 | 3,495 | +23% | 0 | 0 | — |
case-07 | fail→pass | 14,529 | 13,231 | -9% | 1 | 1 | 0% | 2,401 | 2,956 | +23% | 0 | 0 | — |
case-08 | fail→pass | 11,738 | 7,645 | -35% | 1 | 1 | 0% | 1,717 | 2,007 | +17% | 0 | 0 | — |
case-09 | pass→pass | 13,352 | 11,505 | -14% | 1 | 1 | 0% | 2,148 | 2,805 | +31% | 0 | 0 | — |
case-10 | pass→pass | 12,858 | 20,453 | +59% | 1 | 1 | 0% | 2,040 | 4,904 | +140% | 0 | 0 | — |
case-12 | fail→pass | 17,630 | 11,597 | -34% | 1 | 1 | 0% | 2,718 | 2,687 | -1% | 0 | 0 | — |
case-13 | fail→pass | 16,871 | 10,157 | -40% | 1 | 1 | 0% | 2,902 | 2,596 | -11% | 0 | 0 | — |
case-14 | pass→pass | 12,334 | 10,922 | -11% | 1 | 1 | 0% | 1,837 | 2,583 | +41% | 0 | 0 | — |
case-15 | pass→pass | 10,892 | 7,415 | -32% | 1 | 1 | 0% | 1,684 | 1,940 | +15% | 0 | 0 | — |
case-16 | pass→pass | 11,460 | 8,785 | -23% | 1 | 1 | 0% | 1,850 | 2,240 | +21% | 0 | 0 | — |
case-17 | fail→pass | 9,619 | 10,960 | +14% | 1 | 1 | 0% | 1,661 | 2,853 | +72% | 0 | 0 | — |
case-18 | pass→pass | 10,843 | 12,028 | +11% | 1 | 1 | 0% | 1,785 | 2,834 | +59% | 0 | 0 | — |
case-19 | pass→pass | 11,971 | 10,755 | -10% | 1 | 1 | 0% | 1,862 | 2,571 | +38% | 0 | 0 | — |
case-20 | fail→pass | 14,747 | 23,492 | +59% | 1 | 1 | 0% | 2,338 | 3,671 | +57% | 0 | 0 | — |
case-21 | pass→pass | 13,998 | 12,850 | -8% | 1 | 1 | 0% | 2,208 | 3,008 | +36% | 0 | 0 | — |
case-22 | pass→pass | 10,705 | 4,334 | -60% | 1 | 1 | 0% | 1,484 | 1,492 | +1% | 0 | 0 | — |
case-23 | pass→pass | 13,850 | 13,509 | -2% | 1 | 1 | 0% | 2,157 | 2,856 | +32% | 0 | 0 | — |
case-24 | fail→fail | 15,273 | 12,963 | -15% | 1 | 1 | 0% | 2,610 | 2,928 | +12% | 0 | 0 | — |
case-25 | fail→fail | 38,812 | 20,586 | -47% | 1 | 1 | 0% | 8,219 | 4,664 | -43% | 0 | 0 | — |
case-26 | pass→pass | 8,636 | 8,291 | -4% | 1 | 1 | 0% | 1,857 | 2,382 | +28% | 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. 26 cases were attempted. The headline lift of +31 percentage points is the difference between those two pass rates over the 26 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.