Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create short videos and motion clips from a text description using AI video generation.
.claude/skills/holaboss-ai-video-generator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 44% | 0% |
Act as the director who briefs the video model. A video prompt has to carry everything a still image does plus time — what moves, how the camera behaves, and what changes between the first frame and the last. Your job is to turn a one-line request into a shot that reads as deliberate rather than as a wobbling still.
Use Video Generator for short social clips, product motion, b-roll, animated backgrounds, and any request that names a video, clip, animation, or motion. Reach for it whenever the deliverable moves.
Build the shot deliberately rather than describing a picture and hoping it animates. Cover:
Be specific where it matters and silent where it doesn't. Contradictory motion (a static camera that also orbits) is the most common cause of a smeared result.
Return the generated clip when generation is available; otherwise return the finished, ready-to-use prompt. For each, note the aspect ratio and duration it targets and one line on the directorial choice. When delivering variations, label each by the axis it explores.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 13,776 | 12,783 | -7% | 1 | 1 | 0% | 2,026 | 2,132 | +5% | 0 | 0 | — |
case-02 | fail→fail | 14,051 | 10,005 | -29% | 1 | 1 | 0% | 2,020 | 2,236 | +11% | 0 | 0 | — |
case-03 | pass→pass | 12,416 | 12,766 | +3% | 1 | 1 | 0% | 1,676 | 2,246 | +34% | 0 | 0 | — |
case-04 | pass→pass | 15,326 | 12,816 | -16% | 1 | 1 | 0% | 2,438 | 2,570 | +5% | 0 | 0 | — |
case-05 | pass→fail | 8,268 | 11,016 | +33% | 1 | 1 | 0% | 1,225 | 2,122 | +73% | 0 | 0 | — |
case-06 | fail→fail | 9,633 | 9,139 | -5% | 1 | 1 | 0% | 1,247 | 1,997 | +60% | 0 | 0 | — |
case-07 | fail→pass | 7,420 | 9,885 | +33% | 1 | 1 | 0% | 1,125 | 2,020 | +80% | 0 | 0 | — |
case-08 | fail→pass | 8,587 | 10,221 | +19% | 1 | 1 | 0% | 1,427 | 1,992 | +40% | 0 | 0 | — |
case-09 | fail→pass | 10,552 | 14,999 | +42% | 1 | 1 | 0% | 1,494 | 2,596 | +74% | 0 | 0 | — |
case-10 | fail→pass | 8,836 | 15,467 | +75% | 1 | 1 | 0% | 1,491 | 2,595 | +74% | 0 | 0 | — |
case-11 | fail→pass | 11,156 | 11,565 | +4% | 1 | 1 | 0% | 1,612 | 2,328 | +44% | 0 | 0 | — |
case-12 | fail→fail | 19,940 | 16,108 | -19% | 1 | 1 | 0% | 2,067 | 2,564 | +24% | 0 | 0 | — |
case-13 | fail→fail | 7,612 | 8,650 | +14% | 1 | 1 | 0% | 1,130 | 1,890 | +67% | 0 | 0 | — |
case-14 | fail→fail | 5,819 | 8,018 | +38% | 1 | 1 | 0% | 752 | 1,855 | +147% | 0 | 0 | — |
case-15 | pass→pass | 7,917 | 9,660 | +22% | 1 | 1 | 0% | 1,216 | 1,912 | +57% | 0 | 0 | — |
case-16 | pass→pass | 8,048 | 8,920 | +11% | 1 | 1 | 0% | 1,274 | 1,883 | +48% | 0 | 0 | — |
case-17 | fail→fail | 9,263 | 10,021 | +8% | 1 | 1 | 0% | 1,439 | 2,058 | +43% | 0 | 0 | — |
case-18 | fail→pass | 8,299 | 11,910 | +44% | 1 | 1 | 0% | 1,554 | 2,204 | +42% | 0 | 0 | — |
case-19 | fail→fail | 6,250 | 6,973 | +12% | 1 | 1 | 0% | 810 | 1,558 | +92% | 0 | 0 | — |
case-20 | pass→pass | 7,698 | 9,899 | +29% | 1 | 1 | 0% | 1,295 | 1,917 | +48% | 0 | 0 | — |
case-21 | fail→fail | 8,004 | 9,094 | +14% | 1 | 1 | 0% | 872 | 1,833 | +110% | 0 | 0 | — |
case-22 | fail→fail | 5,288 | 8,091 | +53% | 1 | 1 | 0% | 828 | 1,614 | +95% | 0 | 0 | — |
case-23 | pass→pass | 10,271 | 11,411 | +11% | 1 | 1 | 0% | 1,798 | 2,360 | +31% | 0 | 0 | — |
case-24 | fail→pass | 8,904 | 8,422 | -5% | 1 | 1 | 0% | 1,251 | 2,046 | +64% | 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. 24 cases were attempted. The headline lift of +25 percentage points is the difference between those two pass rates over the 24 comparable cases. 2 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.