Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Add or revise scene-bound narration in an active iPolloWork Video Studio project using its selected voice, media actions, timing contract, and final validation.
.claude/skills/devin-axis-ipollowork-video-voiceover/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 142% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -65% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -35% | 0% |
Use this Skill only when narration helps the confirmed brief or the user explicitly requests it. The Video Studio and its media services remain core iPolloWork capabilities and do not depend on this Skill being installed.
voiceover.json, and injected voiceover contract before synthesizing anything.If no valid voice selection exists, continue without narration instead of inventing a voice or blocking visual video work.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,620 | 20,434 | +40% | 1 | 1 | 0% | 1,540 | 2,754 | +79% | 0 | 0 | — |
case-02 | fail→pass | 15,524 | 15,435 | -1% | 1 | 1 | 0% | 1,752 | 2,077 | +19% | 0 | 0 | — |
case-03 | fail→pass | 22,035 | 21,174 | -4% | 1 | 1 | 0% | 1,454 | 3,513 | +142% | 0 | 0 | — |
case-04 | fail→fail | 14,926 | 7,984 | -47% | 1 | 1 | 0% | 1,557 | 756 | -51% | 0 | 0 | — |
case-05 | fail→pass | 12,690 | 8,090 | -36% | 1 | 1 | 0% | 1,196 | 748 | -37% | 0 | 0 | — |
case-06 | fail→pass | 19,985 | 8,230 | -59% | 1 | 1 | 0% | 2,179 | 756 | -65% | 0 | 0 | — |
case-07 | fail→pass | 18,580 | 10,943 | -41% | 1 | 1 | 0% | 1,953 | 1,269 | -35% | 0 | 0 | — |
case-08 | fail→pass | 14,531 | 8,494 | -42% | 1 | 1 | 0% | 1,339 | 830 | -38% | 0 | 0 | — |
case-09 | fail→pass | 14,809 | 7,528 | -49% | 1 | 1 | 0% | 1,455 | 633 | -56% | 0 | 0 | — |
case-10 | fail→fail | 19,713 | 7,830 | -60% | 1 | 1 | 0% | 2,302 | 744 | -68% | 0 | 0 | — |
case-11 | pass→pass | 10,077 | 6,807 | -32% | 1 | 1 | 0% | 771 | 508 | -34% | 0 | 0 | — |
case-12 | fail→pass | 12,615 | 7,856 | -38% | 1 | 1 | 0% | 1,218 | 741 | -39% | 0 | 0 | — |
case-13 | pass→pass | 14,805 | 7,880 | -47% | 1 | 1 | 0% | 1,401 | 649 | -54% | 0 | 0 | — |
case-14 | pass→pass | 14,676 | 9,102 | -38% | 1 | 1 | 0% | 1,494 | 946 | -37% | 0 | 0 | — |
case-15 | pass→pass | 12,437 | 11,829 | -5% | 1 | 1 | 0% | 1,318 | 1,415 | +7% | 0 | 0 | — |
case-16 | fail→pass | 15,067 | 10,390 | -31% | 1 | 1 | 0% | 1,574 | 1,057 | -33% | 0 | 0 | — |
case-17 | fail→pass | 17,400 | 7,700 | -56% | 1 | 1 | 0% | 2,066 | 741 | -64% | 0 | 0 | — |
case-18 | pass→pass | 15,609 | 12,305 | -21% | 1 | 1 | 0% | 1,686 | 1,486 | -12% | 0 | 0 | — |
case-19 | pass→pass | 11,647 | 12,849 | +10% | 1 | 1 | 0% | 1,099 | 1,526 | +39% | 0 | 0 | — |
case-20 | pass→fail | 14,972 | 11,257 | -25% | 1 | 1 | 0% | 1,709 | 1,213 | -29% | 0 | 0 | — |
case-21 | pass→fail | 14,716 | 7,403 | -50% | 1 | 1 | 0% | 1,491 | 640 | -57% | 0 | 0 | — |
case-22 | fail→pass | 15,972 | 7,966 | -50% | 1 | 1 | 0% | 1,614 | 773 | -52% | 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 +41 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 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.