Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a full creator newsletter issue — subject line, preview text, hook, body with a clear takeaway, and a CTA — in the writer's voice, for Substack, beehiiv, ConvertKit, or email. Use when asked to write a newsletter, an email issue, a Substack post, or to turn notes/a topic into a sendable newsletter. Produces a ready-to-send issue with subject-line options and a skimmable structure. Distinct from B2B drip/nurture sequences.
.claude/skills/itamarzand88-newsletter-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 68% | 0% |
<!-- source: newsletter-writer — https://raw.githubusercontent.com/mohitagw15856/pm-claude-skills/main/skills/newsletter-writer/SKILL.md -->
A newsletter is a relationship, not a broadcast. The best issues open a loop in the subject line, reward the open in the first sentence, deliver one clear idea, and make the next step obvious. This skill writes that issue — in your voice, ready to send.
Given a topic, notes, or a piece to adapt, write the full issue anyway. Infer the audience and the single idea; mark invented specifics (assumed — replace). Don't hedge with "in this issue we'll explore…" — get to the value.
Ask for (if not already provided):
5 options, mixing curiosity, specificity, and benefit — each ≤ ~50 characters. Star the recommended one.
One line (~80 chars) that complements (doesn't repeat) the subject.
A 3-bullet "what a 5-second skimmer takes away" — if those bullets don't carry the value, restructure.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→fail | 11,444 | 11,609 | +1% | 1 | 1 | 0% | 2,606 | 3,071 | +18% | 0 | 0 | — |
case-01 | fail→fail | 13,597 | 10,318 | -24% | 1 | 1 | 0% | 2,232 | 2,413 | +8% | 0 | 0 | — |
case-02 | fail→fail | 12,449 | 11,948 | -4% | 1 | 1 | 0% | 2,147 | 2,710 | +26% | 0 | 0 | — |
case-03 | fail→pass | 14,746 | 11,388 | -23% | 1 | 1 | 0% | 2,579 | 2,575 | -0% | 0 | 0 | — |
case-04 | pass→pass | 13,325 | 12,762 | -4% | 1 | 1 | 0% | 2,416 | 3,096 | +28% | 0 | 0 | — |
case-06 | pass→pass | 7,099 | 9,397 | +32% | 1 | 1 | 0% | 1,476 | 2,585 | +75% | 0 | 0 | — |
case-07 | pass→pass | 10,328 | 10,076 | -2% | 1 | 1 | 0% | 1,824 | 2,561 | +40% | 0 | 0 | — |
case-08 | fail→fail | 11,531 | 11,386 | -1% | 1 | 1 | 0% | 2,213 | 2,598 | +17% | 0 | 0 | — |
case-09 | fail→pass | 5,928 | 3,625 | -39% | 1 | 1 | 0% | 1,122 | 1,359 | +21% | 0 | 0 | — |
case-10 | pass→fail | 4,667 | 5,949 | +27% | 1 | 1 | 0% | 906 | 1,660 | +83% | 0 | 0 | — |
case-11 | fail→fail | 11,444 | 11,886 | +4% | 1 | 1 | 0% | 2,015 | 2,850 | +41% | 0 | 0 | — |
case-12 | fail→fail | 9,289 | 9,230 | -1% | 1 | 1 | 0% | 1,726 | 2,289 | +33% | 0 | 0 | — |
case-13 | fail→pass | 11,224 | 9,807 | -13% | 1 | 1 | 0% | 2,158 | 2,481 | +15% | 0 | 0 | — |
case-14 | pass→pass | 14,031 | 10,958 | -22% | 1 | 1 | 0% | 2,505 | 2,614 | +4% | 0 | 0 | — |
case-15 | fail→pass | 12,268 | 10,027 | -18% | 1 | 1 | 0% | 2,420 | 2,473 | +2% | 0 | 0 | — |
case-16 | fail→pass | 6,050 | 7,684 | +27% | 1 | 1 | 0% | 1,219 | 2,050 | +68% | 0 | 0 | — |
case-17 | fail→pass | 20,568 | 12,656 | -38% | 1 | 1 | 0% | 3,332 | 2,756 | -17% | 0 | 0 | — |
case-18 | fail→fail | 10,754 | 8,045 | -25% | 1 | 1 | 0% | 1,904 | 2,267 | +19% | 0 | 0 | — |
case-19 | fail→pass | 8,267 | 9,354 | +13% | 1 | 1 | 0% | 1,445 | 2,426 | +68% | 0 | 0 | — |
case-20 | fail→pass | 10,731 | 9,933 | -7% | 1 | 1 | 0% | 2,143 | 2,539 | +18% | 0 | 0 | — |
case-21 | pass→pass | 9,736 | 9,762 | +0% | 1 | 1 | 0% | 1,948 | 2,772 | +42% | 0 | 0 | — |
case-22 | pass→pass | 8,639 | 9,180 | +6% | 1 | 1 | 0% | 1,748 | 2,320 | +33% | 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 +27 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.