Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Draft blog posts, social media, email newsletters, landing pages, press releases, and case studies with channel-specific formatting and SEO recommendations. Use when writing any marketing content, when you need headline or subject line options, or when adapting a message for a specific platform, audience, and brand voice.
.claude/skills/evolution-foundation-mkt-draft-content/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -33% | 0% |
> If you see unfamiliar placeholders or need to check which tools are connected, see CONNECTORS.md.
Generate marketing content drafts tailored to a specific content type, audience, and brand voice.
User runs /draft-content or asks to draft, write, or create marketing content.
Gather the following from the user. If not provided, ask before proceeding:
For blog posts, landing pages, and other web-facing content:
Present the draft with clear formatting. After the draft, include:
Ask: "Would you like me to revise any section, adjust the tone, or create a variation for a different channel?"
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-14 | pass→pass | 17,400 | 10,914 | -37% | 1 | 1 | 0% | 2,773 | 2,645 | -5% | 0 | 0 | — |
case-01 | fail→pass | 21,270 | 14,475 | -32% | 1 | 1 | 0% | 3,148 | 3,165 | +1% | 0 | 0 | — |
case-02 | fail→fail | 7,961 | 11,966 | +50% | 1 | 1 | 0% | 1,372 | 3,056 | +123% | 0 | 0 | — |
case-03 | fail→pass | 8,782 | 9,833 | +12% | 1 | 1 | 0% | 1,432 | 2,522 | +76% | 0 | 0 | — |
case-04 | pass→pass | 11,711 | 7,002 | -40% | 1 | 1 | 0% | 2,124 | 2,129 | +0% | 0 | 0 | — |
case-05 | fail→pass | 11,374 | 10,517 | -8% | 1 | 1 | 0% | 1,800 | 2,495 | +39% | 0 | 0 | — |
case-06 | fail→pass | 13,581 | 10,052 | -26% | 1 | 1 | 0% | 2,217 | 2,586 | +17% | 0 | 0 | — |
case-07 | pass→fail | 5,097 | 5,930 | +16% | 1 | 1 | 0% | 761 | 1,933 | +154% | 0 | 0 | — |
case-08 | pass→pass | 8,311 | 8,565 | +3% | 1 | 1 | 0% | 1,403 | 2,264 | +61% | 0 | 0 | — |
case-09 | fail→pass | 14,908 | 4,125 | -72% | 1 | 1 | 0% | 2,420 | 1,614 | -33% | 0 | 0 | — |
case-10 | fail→pass | 14,722 | 13,192 | -10% | 1 | 1 | 0% | 2,377 | 3,069 | +29% | 0 | 0 | — |
case-11 | fail→fail | 18,937 | 12,595 | -33% | 1 | 1 | 0% | 2,962 | 2,942 | -1% | 0 | 0 | — |
case-12 | fail→pass | 11,035 | 10,495 | -5% | 1 | 1 | 0% | 1,899 | 2,783 | +47% | 0 | 0 | — |
case-13 | fail→fail | 9,637 | 8,070 | -16% | 1 | 1 | 0% | 1,707 | 2,293 | +34% | 0 | 0 | — |
case-15 | pass→pass | 15,697 | 10,838 | -31% | 1 | 1 | 0% | 2,693 | 2,694 | +0% | 0 | 0 | — |
case-16 | fail→fail | 8,709 | 7,345 | -16% | 1 | 1 | 0% | 1,232 | 2,039 | +66% | 0 | 0 | — |
case-17 | fail→pass | 16,339 | 12,479 | -24% | 1 | 1 | 0% | 2,669 | 2,992 | +12% | 0 | 0 | — |
case-18 | fail→pass | 17,423 | 14,512 | -17% | 1 | 1 | 0% | 2,480 | 3,139 | +27% | 0 | 0 | — |
case-19 | fail→fail | 18,108 | 8,858 | -51% | 1 | 1 | 0% | 2,925 | 2,331 | -20% | 0 | 0 | — |
case-20 | fail→fail | 16,511 | 13,543 | -18% | 1 | 1 | 0% | 2,733 | 3,096 | +13% | 0 | 0 | — |
case-21 | fail→fail | 12,913 | 10,994 | -15% | 1 | 1 | 0% | 2,132 | 2,732 | +28% | 0 | 0 | — |
case-22 | fail→fail | 2,202 | 2,579 | +17% | 1 | 1 | 0% | 271 | 1,301 | +380% | 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. 1 case got worse with the skill loaded, and it is 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.