Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create or update reusable iPolloWork templates through conversation for Design, Slides/native editable PPT, and HyperFrames Video. Use when planning, generating, repairing, validating, or saving a template package, manifest, reusable variables, cover, slide/PPT markers, video composition, or design-system token integration.
.claude/skills/devin-axis-ipollowork-template-generation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -28% | 0% |
Create a reusable template, not a one-off artifact. The product-selected category and server validator are authoritative.
Guide the user in this order: purpose and audience, content structure, reusable variables, visual/design system, surface-specific constraints, generation and validation.
manifest.json, the surface entry, design-tokens.css, local assets when needed, and a 960 x 540 cover.--ipw-* tokens. Do not mix content variables with design tokens.powershellnode .codex/skills/ipollowork-template-generation/scripts/audit-template.mjs <template-directory>
This Skill is guidance for repository agents. Product runtime behavior does not depend on Skill discovery. Authoring System Context and the shared server validator provide the same category rules to every user.
Report the selected surface/category, files changed, reusable variables, design-system coverage, preserved surface invariants, validator result, re-instantiation result, and any remaining warning.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 35,311 | 16,659 | -53% | 1 | 1 | 0% | 6,798 | 965 | -86% | 0 | 0 | — |
case-11 | fail→pass | 9,061 | 10,305 | +14% | 1 | 1 | 0% | 1,366 | 1,386 | +1% | 0 | 0 | — |
case-02 | fail→fail | 14,572 | 14,291 | -2% | 1 | 1 | 0% | 263 | 847 | +222% | 0 | 0 | — |
case-03 | fail→fail | 26,691 | 14,971 | -44% | 1 | 1 | 0% | 4,054 | 867 | -79% | 0 | 0 | — |
case-04 | pass→fail | 16,021 | 7,893 | -51% | 1 | 1 | 0% | 1,922 | 942 | -51% | 0 | 0 | — |
case-05 | pass→pass | 19,765 | 21,048 | +6% | 1 | 1 | 0% | 2,426 | 3,245 | +34% | 0 | 0 | — |
case-06 | pass→pass | 15,771 | 16,243 | +3% | 1 | 1 | 0% | 1,652 | 2,390 | +45% | 0 | 0 | — |
case-07 | fail→pass | 16,196 | 4,091 | -75% | 1 | 1 | 0% | 1,710 | 1,310 | -23% | 0 | 0 | — |
case-08 | fail→pass | 10,189 | 8,816 | -13% | 1 | 1 | 0% | 850 | 1,210 | +42% | 0 | 0 | — |
case-09 | fail→pass | 16,017 | 5,769 | -64% | 1 | 1 | 0% | 1,635 | 1,585 | -3% | 0 | 0 | — |
case-10 | fail→fail | 18,222 | 4,560 | -75% | 1 | 1 | 0% | 2,031 | 1,398 | -31% | 0 | 0 | — |
case-12 | fail→pass | 12,734 | 10,098 | -21% | 1 | 1 | 0% | 2,088 | 1,512 | -28% | 0 | 0 | — |
case-13 | fail→pass | 11,821 | 5,660 | -52% | 1 | 1 | 0% | 1,922 | 1,534 | -20% | 0 | 0 | — |
case-14 | fail→pass | 11,042 | 8,010 | -27% | 1 | 1 | 0% | 1,040 | 1,103 | +6% | 0 | 0 | — |
case-15 | fail→pass | 10,125 | 2,372 | -77% | 1 | 1 | 0% | 821 | 1,003 | +22% | 0 | 0 | — |
case-16 | fail→fail | 10,366 | 3,700 | -64% | 1 | 1 | 0% | 1,513 | 1,178 | -22% | 0 | 0 | — |
case-17 | pass→fail | 8,376 | 4,271 | -49% | 1 | 1 | 0% | 1,358 | 1,273 | -6% | 0 | 0 | — |
case-18 | fail→pass | 11,208 | 5,827 | -48% | 1 | 1 | 0% | 1,767 | 1,572 | -11% | 0 | 0 | — |
case-19 | fail→pass | 13,694 | 9,048 | -34% | 1 | 1 | 0% | 2,238 | 2,094 | -6% | 0 | 0 | — |
case-20 | pass→pass | 12,967 | 5,384 | -58% | 1 | 1 | 0% | 1,879 | 1,383 | -26% | 0 | 0 | — |
case-21 | pass→pass | 12,869 | 3,872 | -70% | 1 | 1 | 0% | 1,903 | 1,179 | -38% | 0 | 0 | — |
case-22 | fail→pass | 8,240 | 1,785 | -78% | 1 | 1 | 0% | 1,636 | 854 | -48% | 0 | 0 | — |
case-23 | fail→fail | 7,874 | 3,090 | -61% | 1 | 1 | 0% | 1,143 | 1,067 | -7% | 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. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +39 percentage points is the difference between those two pass rates over the 20 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.