Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Brainstorm creative content ideas tailored to your niche, audience, and goals.
.claude/skills/holaboss-ai-idea-generator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 65% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 148% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 25% | 0% |
Brainstorm like a creative lead who never ships a blank list: every idea comes with an angle, a hook, and a home. The goal is fresh, relevant, usable concepts the team can turn into real content — not generic prompts anyone could have guessed.
Use Idea Generator to produce content ideas tailored to a niche, audience, and goal: topics, series concepts, hooks, formats, and the gaps worth filling. If those ideas need scheduling into a calendar, hand off to Content Planner; if one needs writing, hand off to Content Writer; if you need to know what's currently rising, pull from Trend Spotter first.
Return a list of ideas, each as a compact unit:
Where useful, add a short Series concepts section grouping ideas into themed runs, and a brief Gaps note on what the audience wants that isn't being made yet.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | pass→pass | 24,033 | 26,908 | +12% | 1 | 1 | 0% | 3,505 | 4,527 | +29% | 0 | 0 | — |
case-03 | pass→fail | 16,960 | 12,316 | -27% | 1 | 1 | 0% | 2,452 | 2,646 | +8% | 0 | 0 | — |
case-04 | fail→pass | 10,658 | 35,177 | +230% | 1 | 1 | 0% | 1,628 | 2,688 | +65% | 0 | 0 | — |
case-01 | pass→pass | 26,594 | 34,374 | +29% | 1 | 1 | 0% | 4,987 | 5,472 | +10% | 0 | 0 | — |
case-05 | fail→pass | 8,143 | 14,591 | +79% | 1 | 1 | 0% | 1,122 | 2,784 | +148% | 0 | 0 | — |
case-06 | pass→pass | 20,545 | 20,814 | +1% | 1 | 1 | 0% | 2,569 | 3,199 | +25% | 0 | 0 | — |
case-07 | fail→pass | 20,340 | 15,061 | -26% | 1 | 1 | 0% | 2,863 | 2,825 | -1% | 0 | 0 | — |
case-08 | fail→pass | 17,399 | 15,432 | -11% | 1 | 1 | 0% | 2,521 | 2,820 | +12% | 0 | 0 | — |
case-09 | fail→fail | 16,613 | 16,109 | -3% | 1 | 1 | 0% | 2,775 | 2,900 | +5% | 0 | 0 | — |
case-10 | fail→fail | 17,431 | 15,627 | -10% | 1 | 1 | 0% | 2,228 | 2,745 | +23% | 0 | 0 | — |
case-11 | fail→fail | 17,521 | 15,768 | -10% | 1 | 1 | 0% | 2,627 | 2,880 | +10% | 0 | 0 | — |
case-12 | fail→pass | 15,688 | 15,708 | +0% | 1 | 1 | 0% | 2,259 | 2,834 | +25% | 0 | 0 | — |
case-13 | fail→pass | 16,118 | 19,208 | +19% | 1 | 1 | 0% | 2,502 | 3,034 | +21% | 0 | 0 | — |
case-14 | fail→fail | 15,790 | 13,220 | -16% | 1 | 1 | 0% | 2,420 | 2,455 | +1% | 0 | 0 | — |
case-15 | fail→pass | 18,573 | 16,601 | -11% | 1 | 1 | 0% | 2,402 | 3,057 | +27% | 0 | 0 | — |
case-16 | fail→pass | 19,057 | 15,919 | -16% | 1 | 1 | 0% | 2,351 | 2,900 | +23% | 0 | 0 | — |
case-17 | pass→pass | 15,832 | 17,305 | +9% | 1 | 1 | 0% | 2,750 | 3,154 | +15% | 0 | 0 | — |
case-18 | fail→pass | 20,306 | 21,360 | +5% | 1 | 1 | 0% | 2,634 | 3,408 | +29% | 0 | 0 | — |
case-19 | fail→fail | 17,953 | 20,038 | +12% | 1 | 1 | 0% | 2,617 | 3,334 | +27% | 0 | 0 | — |
case-20 | fail→fail | 21,146 | 16,580 | -22% | 1 | 1 | 0% | 2,701 | 2,722 | +1% | 0 | 0 | — |
case-21 | pass→pass | 18,138 | 13,524 | -25% | 1 | 1 | 0% | 2,439 | 2,308 | -5% | 0 | 0 | — |
case-22 | fail→fail | 19,906 | 14,460 | -27% | 1 | 1 | 0% | 2,657 | 2,312 | -13% | 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. 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.