Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate high-quality written content for any platform with AI-powered drafting and editing.
.claude/skills/holaboss-ai-content-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 32% | 0% |
Write as a versatile staff writer who can move between a LinkedIn thought-leadership post, a product launch email, and an Instagram caption without losing the client's voice. Every draft should read like it was written by someone who knows the brand, the platform, and the reader — not a generic content mill.
Reach for Content Writer whenever the task is to produce net-new written copy or substantially rework an existing draft: blog posts, landing pages, ad copy, email sequences, social captions, scripts, newsletters. If the job is only to shift register without changing substance, hand off to Tone Adapter; if it's to condense, use Summarizer.
Pin down four things, asking only if they're genuinely missing:
Return the finished copy ready to publish. Lead with the deliverable itself. If you made assumptions about voice, audience, or platform, note them in one short line after the copy — not before it. For multiple variations, separate them clearly with a one-line label describing each one's angle.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,712 | 16,789 | +73% | 1 | 1 | 0% | 1,591 | 1,780 | +12% | 0 | 0 | — |
case-02 | fail→pass | 12,799 | 8,433 | -34% | 1 | 1 | 0% | 1,925 | 1,813 | -6% | 0 | 0 | — |
case-03 | fail→pass | 11,771 | 8,729 | -26% | 1 | 1 | 0% | 1,666 | 1,785 | +7% | 0 | 0 | — |
case-04 | fail→fail | 4,585 | 3,790 | -17% | 1 | 1 | 0% | 922 | 1,204 | +31% | 0 | 0 | — |
case-05 | pass→pass | 5,831 | 7,142 | +22% | 1 | 1 | 0% | 1,200 | 1,777 | +48% | 0 | 0 | — |
case-06 | pass→pass | 3,644 | 2,623 | -28% | 1 | 1 | 0% | 503 | 862 | +71% | 0 | 0 | — |
case-07 | fail→fail | 10,432 | 9,789 | -6% | 1 | 1 | 0% | 1,482 | 1,896 | +28% | 0 | 0 | — |
case-08 | pass→pass | 10,001 | 5,988 | -40% | 1 | 1 | 0% | 1,529 | 1,695 | +11% | 0 | 0 | — |
case-09 | pass→pass | 7,574 | 7,996 | +6% | 1 | 1 | 0% | 1,176 | 1,559 | +33% | 0 | 0 | — |
case-10 | fail→fail | 6,068 | 8,280 | +36% | 1 | 1 | 0% | 961 | 1,874 | +95% | 0 | 0 | — |
case-11 | pass→pass | 10,264 | 10,417 | +1% | 1 | 1 | 0% | 1,553 | 1,934 | +25% | 0 | 0 | — |
case-12 | fail→fail | 9,243 | 8,539 | -8% | 1 | 1 | 0% | 1,542 | 1,697 | +10% | 0 | 0 | — |
case-13 | fail→fail | 10,397 | 8,821 | -15% | 1 | 1 | 0% | 1,604 | 1,820 | +13% | 0 | 0 | — |
case-14 | fail→pass | 9,523 | 6,462 | -32% | 1 | 1 | 0% | 1,499 | 1,537 | +3% | 0 | 0 | — |
case-15 | fail→fail | 12,614 | 10,314 | -18% | 1 | 1 | 0% | 1,681 | 1,825 | +9% | 0 | 0 | — |
case-16 | fail→pass | 9,431 | 7,046 | -25% | 1 | 1 | 0% | 1,227 | 1,621 | +32% | 0 | 0 | — |
case-17 | pass→pass | 11,085 | 10,497 | -5% | 1 | 1 | 0% | 1,631 | 2,114 | +30% | 0 | 0 | — |
case-18 | fail→pass | 13,095 | 9,214 | -30% | 1 | 1 | 0% | 1,896 | 1,951 | +3% | 0 | 0 | — |
case-19 | pass→pass | 8,493 | 8,652 | +2% | 1 | 1 | 0% | 1,297 | 1,843 | +42% | 0 | 0 | — |
case-20 | fail→fail | 8,165 | 7,824 | -4% | 1 | 1 | 0% | 1,121 | 1,611 | +44% | 0 | 0 | — |
case-21 | fail→pass | 4,741 | 5,143 | +8% | 1 | 1 | 0% | 739 | 1,213 | +64% | 0 | 0 | — |
case-22 | fail→pass | 8,976 | 6,837 | -24% | 1 | 1 | 0% | 1,344 | 1,576 | +17% | 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.