Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a specific, non-generic cover letter that connects your evidence to the role. Use when asked to write a cover letter, an application letter, or a note to accompany a resume. Produces a tight 3–4 paragraph letter — a real hook, two evidence paragraphs mapping your proof to the job's needs, and a confident close — tailored to the company, ready to export as a designed PDF.
.claude/skills/mohitagw15856-cover-letter/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-14 | ✓→✗ | ▼ Worse | 24% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 35% | 0% |
Most cover letters are throat-clearing the reader skips. A good one does one job: connect your specific evidence to this company's specific need, in a voice that sounds like a person. This skill writes a tight, tailored letter — no "I am writing to apply for…", no restating the resume — that earns the read.
Ask for these only if they aren't already provided:
A 3–4 paragraph letter (≈250–350 words):
> Name] · email] · phone] · date] > Dear hiring manager name, or "Hiring Team" if unknown],
Hook (1 short para) — open with a specific reason you're writing to them — a genuine connection to their product/mission/moment — and the role you want. No "I am writing to apply."
Evidence (1–2 paras) — the heart: take the role's top 2–3 needs and show, with a concrete result each, that you've done it. Map your proof to their problem; don't recap the resume — interpret it.
Close (1 short para) — what you'd bring, an honest note of enthusiasm, and a forward-looking line ("I'd love to talk about…"). Confident, not desperate.
> Sincerely, Name]
Voice note (for the user): keep it human — contractions, active voice, no thesaurus words. Read it aloud; if it sounds like a robot, cut.
Modern cover-letter practice — specific hook, evidence-to-need mapping, human voice.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 10,787 | 69,280 | +542% | 1 | 1 | 0% | 1,789 | 2,314 | +29% | 0 | 0 | — |
case-02 | fail→pass | 12,549 | 9,831 | -22% | 1 | 1 | 0% | 1,927 | 2,414 | +25% | 0 | 0 | — |
case-03 | fail→pass | 18,652 | 16,852 | -10% | 1 | 1 | 0% | 1,878 | 2,308 | +23% | 0 | 0 | — |
case-04 | pass→pass | 10,032 | 8,660 | -14% | 1 | 1 | 0% | 1,536 | 2,070 | +35% | 0 | 0 | — |
case-05 | pass→pass | 11,921 | 10,186 | -15% | 1 | 1 | 0% | 1,874 | 2,329 | +24% | 0 | 0 | — |
case-18 | pass→pass | 14,826 | 8,922 | -40% | 1 | 1 | 0% | 1,770 | 2,173 | +23% | 0 | 0 | — |
case-06 | pass→pass | 9,169 | 9,405 | +3% | 1 | 1 | 0% | 1,616 | 2,190 | +36% | 0 | 0 | — |
case-07 | pass→pass | 10,433 | 8,794 | -16% | 1 | 1 | 0% | 1,535 | 2,141 | +39% | 0 | 0 | — |
case-08 | pass→pass | 72,020 | 9,730 | -86% | 1 | 1 | 0% | 1,694 | 2,030 | +20% | 0 | 0 | — |
case-09 | pass→pass | 11,807 | 10,036 | -15% | 1 | 1 | 0% | 1,789 | 2,159 | +21% | 0 | 0 | — |
case-19 | pass→pass | 10,828 | 10,470 | -3% | 1 | 1 | 0% | 1,794 | 2,259 | +26% | 0 | 0 | — |
case-10 | fail→pass | 13,593 | 10,694 | -21% | 1 | 1 | 0% | 2,018 | 2,305 | +14% | 0 | 0 | — |
case-11 | pass→pass | 12,224 | 8,399 | -31% | 1 | 1 | 0% | 1,841 | 1,975 | +7% | 0 | 0 | — |
case-12 | pass→pass | 10,024 | 9,058 | -10% | 1 | 1 | 0% | 1,635 | 2,158 | +32% | 0 | 0 | — |
case-13 | pass→pass | 11,266 | 9,987 | -11% | 1 | 1 | 0% | 1,699 | 2,146 | +26% | 0 | 0 | — |
case-20 | pass→pass | 8,257 | 9,363 | +13% | 1 | 1 | 0% | 1,511 | 2,366 | +57% | 0 | 0 | — |
case-14 | pass→fail | 10,967 | 8,437 | -23% | 1 | 1 | 0% | 1,607 | 1,992 | +24% | 0 | 0 | — |
case-15 | pass→pass | 10,387 | 9,782 | -6% | 1 | 1 | 0% | 1,691 | 2,242 | +33% | 0 | 0 | — |
case-16 | fail→fail | 8,548 | 9,593 | +12% | 1 | 1 | 0% | 1,637 | 2,175 | +33% | 0 | 0 | — |
case-17 | fail→fail | 11,719 | 9,620 | -18% | 1 | 1 | 0% | 1,771 | 2,222 | +25% | 0 | 0 | — |
case-21 | pass→pass | 14,721 | 11,449 | -22% | 1 | 1 | 0% | 2,365 | 2,849 | +20% | 0 | 0 | — |
case-22 | pass→pass | 6,207 | 5,524 | -11% | 1 | 1 | 0% | 1,101 | 1,467 | +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 +9 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.