Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Turn real customer reviews, quotes, and case-study results into polished social-proof videos built in HyperFrames -- single-testimonial spotlights, review-wall montages, and stat-driven proof reels with captions and voiceover, cut for vertical, square, and wide. The highest-converting content most businesses never make.
.claude/skills/onewave-ai-hyperframes-testimonial-builder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 136% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 13% | 0% |
Nothing sells like a customer saying it. This skill takes the proof a business already has -- Google reviews, testimonial emails, case-study numbers, NPS comments -- and directs it into renderable HyperFrames video. It owns the selection and the story; for composition mechanics, captions, transitions, and TTS, invoke the hyperframes skill (hyperframes-cli for init/lint/preview/render, hyperframes-media for voiceover).
Take whatever exists: pasted reviews, a review export, customer-review-aggregator output, case-study docs, or a profile URL to pull from. For each candidate quote capture the text verbatim, the name/attribution available, the specific result mentioned, and the source.
Select for specificity: "They saved us $40K and two weeks" outperforms "Great company, highly recommend" every time. Rank candidates by concrete detail, emotional arc (skeptic-to-believer beats always-happy), and relevance to the audience the video targets.
claude-design-system-architect or design-style-installer token set if present); no purple, no emoji.Platform cuts (9:16, 1:1, 16:9), a thumbnail frame, and the caption text for the post. Note where each cut runs best: review wall for the Google Business Profile and Instagram, spotlight for the website hero and proposals, proof reel for paid.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 16,771 | 15,365 | -8% | 1 | 1 | 0% | 2,824 | 3,500 | +24% | 0 | 0 | — |
case-02 | fail→pass | 12,955 | 18,929 | +46% | 1 | 1 | 0% | 1,904 | 4,499 | +136% | 0 | 0 | — |
case-03 | fail→fail | 16,342 | 20,627 | +26% | 1 | 1 | 0% | 2,565 | 4,981 | +94% | 0 | 0 | — |
case-04 | pass→pass | 27,130 | 20,729 | -24% | 1 | 1 | 0% | 4,524 | 4,117 | -9% | 0 | 0 | — |
case-05 | pass→pass | 35,179 | 27,873 | -21% | 1 | 1 | 0% | 4,800 | 4,959 | +3% | 0 | 0 | — |
case-06 | pass→pass | 18,801 | 20,032 | +7% | 1 | 1 | 0% | 3,137 | 4,143 | +32% | 0 | 0 | — |
case-07 | fail→fail | 8,162 | 9,831 | +20% | 1 | 1 | 0% | 1,163 | 2,279 | +96% | 0 | 0 | — |
case-08 | fail→pass | 11,629 | 11,907 | +2% | 1 | 1 | 0% | 1,719 | 2,664 | +55% | 0 | 0 | — |
case-09 | pass→pass | 10,159 | 7,912 | -22% | 1 | 1 | 0% | 1,483 | 2,035 | +37% | 0 | 0 | — |
case-10 | pass→pass | 10,938 | 3,612 | -67% | 1 | 1 | 0% | 1,442 | 1,303 | -10% | 0 | 0 | — |
case-11 | fail→pass | 10,606 | 6,833 | -36% | 1 | 1 | 0% | 1,585 | 1,796 | +13% | 0 | 0 | — |
case-12 | fail→fail | 10,751 | 14,035 | +31% | 1 | 1 | 0% | 1,898 | 3,154 | +66% | 0 | 0 | — |
case-13 | fail→fail | 13,296 | 15,886 | +19% | 1 | 1 | 0% | 2,492 | 3,887 | +56% | 0 | 0 | — |
case-14 | fail→pass | 11,510 | 8,263 | -28% | 1 | 1 | 0% | 1,964 | 2,224 | +13% | 0 | 0 | — |
case-15 | pass→pass | 10,622 | 6,733 | -37% | 1 | 1 | 0% | 1,887 | 1,978 | +5% | 0 | 0 | — |
case-16 | pass→pass | 11,693 | 6,641 | -43% | 1 | 1 | 0% | 1,856 | 1,836 | -1% | 0 | 0 | — |
case-17 | pass→pass | 12,771 | 8,775 | -31% | 1 | 1 | 0% | 1,961 | 2,116 | +8% | 0 | 0 | — |
case-18 | fail→pass | 10,008 | 3,832 | -62% | 1 | 1 | 0% | 1,351 | 1,391 | +3% | 0 | 0 | — |
case-19 | pass→pass | 7,403 | 4,932 | -33% | 1 | 1 | 0% | 1,028 | 1,616 | +57% | 0 | 0 | — |
case-20 | pass→pass | 10,928 | 5,652 | -48% | 1 | 1 | 0% | 1,688 | 1,579 | -6% | 0 | 0 | — |
case-21 | fail→pass | 9,722 | 4,070 | -58% | 1 | 1 | 0% | 1,625 | 1,318 | -19% | 0 | 0 | — |
case-22 | fail→pass | 11,838 | 3,636 | -69% | 1 | 1 | 0% | 2,212 | 1,457 | -34% | 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 +36 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.