Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a cold sales/B2B outreach email that earns a reply. Use when asked to write a cold email, a sales outreach email, a prospecting email, or a cold email sequence to a business prospect. Produces a short, personalised email — subject, a relevant opener, one clear value-led ask, and a low-friction CTA — plus 2 follow-ups, written to be replied to, not deleted.
.claude/skills/mohitagw15856-cold-email/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 83 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -2% | 0% |
Cold email works when it's short, clearly about them, and asks for one small thing. Most cold email fails because it's a feature dump that's all about the sender. This skill writes a tight, personalised email built on a real trigger or relevance hook, with a single low-friction ask — plus the follow-ups that actually drive most replies. (For job-search / networking outreach, use outreach-message; this is B2B sales prospecting.)
Ask for these only if they aren't already provided:
Subject lines — 3 options, short (≤6 words), specific, no clickbait or "Quick question."
The email (≤120 words):
Follow-ups — 2 short ones (send ~3–4 days apart): a value-add nudge (a resource/insight, not "just bumping this") and a graceful breakup email ("I'll close the loop — want me to circle back next quarter?"). Most replies come from these.
Note — what to personalise per prospect (the one line that proves it isn't a blast), and the one metric to watch (reply rate, not open rate).
B2B cold-email practice — relevance/trigger-led openers, one-outcome value, single low-friction ask, value-adding follow-up cadence.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→pass | 10,935 | 7,779 | -29% | 1 | 1 | 0% | 1,985 | 2,321 | +17% | 0 | 0 | — |
case-01 | fail→pass | 11,816 | 7,793 | -34% | 1 | 1 | 0% | 2,232 | 2,243 | +0% | 0 | 0 | — |
case-03 | fail→pass | 11,580 | 7,344 | -37% | 1 | 1 | 0% | 2,246 | 2,317 | +3% | 0 | 0 | — |
case-04 | fail→pass | 10,345 | 8,668 | -16% | 1 | 1 | 0% | 1,906 | 2,584 | +36% | 0 | 0 | — |
case-05 | pass→pass | 9,157 | 9,721 | +6% | 1 | 1 | 0% | 1,593 | 2,804 | +76% | 0 | 0 | — |
case-06 | pass→pass | 9,132 | 8,744 | -4% | 1 | 1 | 0% | 1,598 | 2,214 | +39% | 0 | 0 | — |
case-07 | pass→pass | 8,297 | 8,715 | +5% | 1 | 1 | 0% | 1,587 | 2,191 | +38% | 0 | 0 | — |
case-08 | fail→fail | 10,025 | 8,033 | -20% | 1 | 1 | 0% | 1,948 | 2,382 | +22% | 0 | 0 | — |
case-09 | fail→fail | 5,573 | 8,125 | +46% | 1 | 1 | 0% | 1,028 | 2,319 | +126% | 0 | 0 | — |
case-10 | fail→pass | 13,369 | 8,997 | -33% | 1 | 1 | 0% | 2,461 | 2,418 | -2% | 0 | 0 | — |
case-11 | fail→pass | 8,466 | 8,935 | +6% | 1 | 1 | 0% | 1,543 | 2,300 | +49% | 0 | 0 | — |
case-12 | fail→fail | 5,657 | 6,940 | +23% | 1 | 1 | 0% | 906 | 2,049 | +126% | 0 | 0 | — |
case-13 | fail→pass | 9,867 | 7,632 | -23% | 1 | 1 | 0% | 1,775 | 2,264 | +28% | 0 | 0 | — |
case-14 | fail→pass | 9,101 | 12,786 | +40% | 1 | 1 | 0% | 1,507 | 2,867 | +90% | 0 | 0 | — |
case-15 | fail→fail | 16,103 | 10,422 | -35% | 1 | 1 | 0% | 2,755 | 2,811 | +2% | 0 | 0 | — |
case-16 | fail→pass | 4,772 | 9,064 | +90% | 1 | 1 | 0% | 693 | 2,361 | +241% | 0 | 0 | — |
case-17 | fail→fail | 12,111 | 9,754 | -19% | 1 | 1 | 0% | 2,294 | 2,528 | +10% | 0 | 0 | — |
case-18 | pass→pass | 14,268 | 8,250 | -42% | 1 | 1 | 0% | 2,541 | 2,215 | -13% | 0 | 0 | — |
case-19 | fail→pass | 1,991 | 10,936 | +449% | 1 | 1 | 0% | 369 | 2,754 | +646% | 0 | 0 | — |
case-20 | pass→pass | 6,842 | 8,329 | +22% | 1 | 1 | 0% | 1,036 | 2,165 | +109% | 0 | 0 | — |
case-21 | fail→pass | 13,329 | 7,330 | -45% | 1 | 1 | 0% | 2,253 | 2,098 | -7% | 0 | 0 | — |
case-22 | fail→pass | 13,757 | 8,320 | -40% | 1 | 1 | 0% | 2,411 | 2,246 | -7% | 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 +55 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.