Install any skill in seconds. Free to start, no credit card required.
Get Started Free →When the user wants to write an email to an investor — cold outreach, warm intro request, follow-up after a meeting, monthly investor update, or thank-you note. Also activates for "intro email", "investor email", "follow up with VC", or "investor update".
.claude/skills/mkurman-fundraising-email/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 68% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 39% | 0% |
From startup-context: company one-liner, stage, key traction metrics, fundraising status, and any notable social proof (investors, customers, press).
From the user: email type, recipient (investor name and firm), prior relationship context, and desired outcome (meeting, intro, materials review).
.agents/startup-context.md.**To:** [Investor Name]
**Subject:** [Subject line]
[Email body]For investor updates, output a longer structured email with sections (see type 4 below).
Goal: Get a 30-minute meeting. Length: 5-7 sentences, under 150 words.
[Specific hook] — [Company Name] (e.g., "$40K MRR in 6 months, AI contract review — Lexara").Goal: Make it effortless for your connector to intro you. Two parts:
Goal: Maintain momentum, deliver materials, set next steps. Length: 4-8 sentences.
Goal: Keep current and prospective investors informed. Length: 300-500 words.
Goal: Maintain the relationship. Length: 3-5 sentences.
pitch-deck — the email drives the meeting; the deck carries the meetinginvestor-research — research determines who to email and what personalization to useExample prompt: "Write a cold email to Sarah Chen at Greylock. We're an AI code review startup with $55K MRR."
Good output:
> Subject: $55K MRR — AI code review, 4 enterprise customers — CodeLens > > Hi Sarah, > > Your investment in Developer Infrastructure at Greylock — particularly the thesis in your "Next Wave of DevTools" post — is why I wanted to reach out to you specifically. > > I'm the CEO of CodeLens. We automate code review for enterprise engineering teams using LLMs, catching security vulnerabilities and logic errors that slip past existing tools. > > We launched 5 months ago: $55K MRR, 4 enterprise customers (two Fortune 500), growing 30% MoM. We're raising a $3M seed round. > > Would you have 30 minutes this week or next?
Example prompt: "Help me write this month's investor update."
Good output approach: Pull metrics from startup context, ask the founder for this month's highlights and challenges, then produce the structured update. Flag any missing or stale KPIs.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | fail→fail | 11,974 | 9,171 | -23% | 1 | 1 | 0% | 1,652 | 3,088 | +87% | 0 | 0 | — |
case-01 | fail→fail | 5,561 | 7,450 | +34% | 1 | 1 | 0% | 788 | 2,744 | +248% | 0 | 0 | — |
case-02 | fail→pass | 12,615 | 9,691 | -23% | 1 | 1 | 0% | 1,959 | 3,047 | +56% | 0 | 0 | — |
case-03 | fail→pass | 13,843 | 10,628 | -23% | 1 | 1 | 0% | 2,135 | 3,209 | +50% | 0 | 0 | — |
case-04 | fail→pass | 11,174 | 6,115 | -45% | 1 | 1 | 0% | 1,678 | 2,620 | +56% | 0 | 0 | — |
case-05 | fail→pass | 12,168 | 7,272 | -40% | 1 | 1 | 0% | 1,650 | 2,769 | +68% | 0 | 0 | — |
case-06 | pass→pass | 9,505 | 8,887 | -7% | 1 | 1 | 0% | 1,413 | 2,943 | +108% | 0 | 0 | — |
case-07 | pass→pass | 10,005 | 7,516 | -25% | 1 | 1 | 0% | 1,446 | 2,573 | +78% | 0 | 0 | — |
case-08 | pass→pass | 11,701 | 10,771 | -8% | 1 | 1 | 0% | 1,700 | 3,113 | +83% | 0 | 0 | — |
case-09 | pass→pass | 9,569 | 7,269 | -24% | 1 | 1 | 0% | 1,461 | 2,646 | +81% | 0 | 0 | — |
case-10 | fail→fail | 8,065 | 6,598 | -18% | 1 | 1 | 0% | 1,230 | 2,584 | +110% | 0 | 0 | — |
case-11 | fail→pass | 11,174 | 6,279 | -44% | 1 | 1 | 0% | 1,786 | 2,489 | +39% | 0 | 0 | — |
case-12 | fail→pass | 10,784 | 4,470 | -59% | 1 | 1 | 0% | 1,640 | 2,181 | +33% | 0 | 0 | — |
case-13 | pass→pass | 13,268 | 9,799 | -26% | 1 | 1 | 0% | 1,838 | 3,012 | +64% | 0 | 0 | — |
case-14 | pass→pass | 13,456 | 8,363 | -38% | 1 | 1 | 0% | 2,016 | 2,875 | +43% | 0 | 0 | — |
case-15 | pass→pass | 13,437 | 10,301 | -23% | 1 | 1 | 0% | 2,022 | 3,019 | +49% | 0 | 0 | — |
case-17 | fail→pass | 9,984 | 6,942 | -30% | 1 | 1 | 0% | 1,450 | 2,556 | +76% | 0 | 0 | — |
case-18 | pass→pass | 12,719 | 10,741 | -16% | 1 | 1 | 0% | 1,900 | 3,084 | +62% | 0 | 0 | — |
case-19 | pass→pass | 10,762 | 7,807 | -27% | 1 | 1 | 0% | 1,651 | 2,861 | +73% | 0 | 0 | — |
case-20 | fail→fail | 10,852 | 6,916 | -36% | 1 | 1 | 0% | 1,715 | 2,558 | +49% | 0 | 0 | — |
case-21 | pass→pass | 11,089 | 9,238 | -17% | 1 | 1 | 0% | 1,627 | 2,987 | +84% | 0 | 0 | — |
case-22 | fail→pass | 13,203 | 8,696 | -34% | 1 | 1 | 0% | 1,989 | 2,869 | +44% | 0 | 0 | — |
case-23 | pass→pass | 11,528 | 10,111 | -12% | 1 | 1 | 0% | 1,819 | 3,028 | +66% | 0 | 0 | — |
case-24 | pass→pass | 13,146 | 9,305 | -29% | 1 | 1 | 0% | 1,910 | 2,896 | +52% | 0 | 0 | — |
case-25 | pass→pass | 11,586 | 10,139 | -12% | 1 | 1 | 0% | 1,691 | 2,943 | +74% | 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. 25 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 25 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.