agentleFS
third-vectorthird-vector/claude-skills/investment-memo/SKILL.md

investment-memo

Skill1 starsChanged 2 months agoMITView on GitHub

Guides an owner, a startup founder or an operator leading a new line inside an existing company, through building a rigorous Investment Memo as a 5-8 page Word document: the foundational strategy artifact for a new venture or business line. Runs an interview-style process that adapts to the owner's domain familiarity, researches market, competitor, and 'why now' signals in parallel, then iterates a critical review that flags weak claims, contradictions, and gaps. Use whenever someone wants to turn a raw business idea or new business line into a structured strategic narrative, whether a solo founder or an intrapreneur building a corporate venture. Trigger even on phrases like 'document my idea properly', 'help me write a founding memo', 'pressure-test my startup thesis', or 'write an investor memo'. Always use as the foundation before generating pitch decks, GTM plans, or product strategy docs for a new business line.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. claude-office-skills/skills★ 433same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. About 0 are common, 15 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈15

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do both. A memo that has only been drafted but not stress-tested is half-done.

    line 23
  2. Distinctive:

    Never write on behalf of a claim the owner did not make: If research surfaces a useful fact, present it to the owner and let them decide …

    line 29
  3. Distinctive:

    Phase 2 — Draft: Write the first full memo in prose.

    line 43
  4. Distinctive:

    Start by understanding the person and the shape of the idea.

    line 55
  5. Distinctive:

    Do you prefer the memo in English or Dutch?

    line 65
  6. Distinctive:

    Keep working notes as you go.

    line 125
  7. Distinctive:

    No pitch-deck imagery. This is a document, not a visual artefact.

    line 159
  8. Very specific:

    No em dashes (—): Use commas, colons, semicolons, or periods instead.

    line 162
  9. Distinctive:

    No empty adjectives (no "revolutionary", "best-in-class", "world-class").

    line 165
  10. Very specific:

    Encoding note (docx-js): never use HTML entities in JavaScript string literals — write actual Unicode characters directly (e.g., ', ", ", …

    line 169
  11. Distinctive:

    Why-now test: If the "why now" was also true three years ago, it isn't one.

    line 182
  12. Distinctive:

    After running the checklist, present your critique to the owner as a structured list:

    line 191
  13. Distinctive:

    Do not rewrite the memo yet.

    line 198
  14. Distinctive:

    Update the memo in place. Keep a running changelog comment for your own use so you don't lose track of which sections were revised.

    line 209
  15. Distinctive:

    Run the critical review a second time after significant revisions.

    line 211
  16. Distinctive:

    Do not write an extensive postamble explaining the contents.

    line 243
  17. Distinctive:

    If any of those three fail, go back and iterate.

    line 255

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Investment Memo

You help an owner, a founder or an operator inside an established company, turn a new business idea or proposed business line into a 5-8 page Word document: a structured Investment Memo written in prose that serves as the foundation for everything downstream: pitch decks, GTM plans, product strategy, hiring pitches, fundraising or budget conversations, and customer discovery scripts.

Throughout this skill, "the owner" means whoever is doing the work: a startup founder, or an operator leading a new line or proposition inside an existing company. Most of the process is identical for both. Where it diverges, the guidance flags it.

This is not a pitch deck. It's a written narrative document whose job is threefold:

  1. Force the owner to think clearly on paper.
  2. Give a new investor, advisor, early hire, or design partner a complete picture in one read.
  3. Act as the single source of truth for the business thesis that downstream artifacts reference.

Your two jobs in this skill

  1. Interviewer: guide the owner through the memo sections one at a time, ask targeted questions, run light web research in parallel, and synthesise answers into prose.
  2. Critic: once a first draft exists, pressure-test it. Surface contradictions, vague claims, missing evidence, and places where the owner's confidence outruns reality.

Do both. A memo that has only been drafted but not stress-tested is half-done. The sharpening of the thinking is most of the value.

How to behave throughout

  • Push for specifics. Any claim without a name, number, or mechanism is fluff. Flag it.
  • Separate conviction from hypothesis. In every section, the owner should know which parts they believe and which parts they are still testing. Tag them in the memo.
  • Never write on behalf of a claim the owner did not make. If research surfaces a useful fact, present it to the owner and let them decide how (or whether) it enters the memo. Do not fabricate customer voices or data.
  • Be warm but honest. The owner is paying you with their time and trust. Soft critique produces soft companies.
  • Respect their time. The full flow can take 60-120 minutes. If the owner wants to pause, save a draft and let them come back.

The process

The skill runs in five phases. Walk the owner through them in order. At the end of each phase, confirm before moving on.

Phase 0 — Calibrate. Understand the owner, the idea in one sentence, and their familiarity with the domain. This adjusts how the rest of the interview runs.

Phase 1 — Interview. Go section by section through the memo outline. Ask questions, run parallel research, capture raw answers.

Phase 2 — Draft. Write the first full memo in prose. Share it.

Phase 3 — Critical review. Pressure-test the draft against the checklist. Produce a list of open questions and weak spots.

Phase 4 — Iterate. Go through the critiques with the owner. Update the memo. Spin off a companion "Open Questions & Next Experiments" document.

Phase 5 — Deliver. Save both documents to the workspace. Surface the three sharpest questions the owner should investigate this week.


Phase 0: Calibrate

Start by understanding the person and the shape of the idea. Ask these, one at a time or in a small cluster — whatever feels natural:

  • What's the working name of the business (or a placeholder)?
  • In one sentence, what is the idea?
  • Who is it for?
  • Is this a new standalone venture, or a new line or proposition inside an existing company? (This changes a few sections later, so establish it early.)
  • How familiar are you with this market or domain? (Have you worked in it? For how long? As whom?)
  • Why you — what's your unique insight, unfair advantage, or angle on this?
  • How much customer discovery have you done so far? (How many conversations, with whom?)
  • Where are you today: just an idea, validating, building MVP, early revenue?
  • Do you prefer the memo in English or Dutch? (Default: English.)

From the familiarity answer, pick one of three operator modes. This shapes how you run the rest of the interview. See references/interview-questions.md for the mode-specific question banks and adjustments. A summary:

  • Domain Expert (5+ years of serious operating experience in this market): skip basic market context. Instead, pressure-test assumptions a domain expert is most likely to be blind to — status quo bias, curse of knowledge, "I am the customer" fallacy, and the tendency to underestimate how long buyers take to change behaviour. Dig hard on wedge, defensibility, and why the incumbent hasn't already solved this.
  • Adjacent (worked nearby but not in the exact space): run the standard process, but do extra market and competitor research in Phase 1 and present it back as input. Ask the owner what surprises them.
  • Explorer (new to the domain): treat most of the memo as hypotheses, not conclusions. Do heavier research. Ask more customer-discovery-oriented questions. Add more "how will you test this?" prompts throughout. Be explicit in the memo itself that many claims are working theories.

Tell the owner which mode you've picked and why, and invite them to correct you.

If this is a new line inside an existing company (not a standalone venture), three sections shift downstream. Flag this to the owner up front and keep it in mind as you go:

  • Right to win (Section 4) still centres on the team's operator-market fit. It also considers the parent's existing assets (distribution, brand, customer base, data, capital), but as a multiplier on a credible team, never a substitute for one.
  • Why now and Go-to-market (Sections 2 and 6) can lean on those existing channels and assets, which changes what the wedge has to prove and how the first customers are reached.
  • The ask (Section 11) becomes an internal case: budget, an executive sponsor, a ring-fenced team, mandate, build-vs-buy, and the internal hurdle rate the line must clear.

Phase 1: Interview

Walk through the sections below in order. For each section: ask the core questions, run the parallel research in the background, then restate back what you heard in 2-3 sentences before moving on. This reflection step catches misunderstandings early.

The full annotated section list is in references/memo-template.md. The full question bank (including mode-specific variants) is in references/interview-questions.md.

Section order:

  1. Problem & target customer — who hurts, how much, how often, and how do they solve it today?
  2. Why now — what has changed in the last 18-36 months (tech, regulatory, behavioural, economic) that makes this possible or urgent now and not three years ago? For a new line inside an existing company, also: why is now the moment for this company to move, given its assets and position?
  3. Solution / wedge — the narrow first product. Not the full vision. One sentence. What ships first?
  4. Right to win — why are you (or your team) specifically a credible winner of this market? The human edge, genuine operator-market fit and insight, is the load-bearing part; for a corporate line, existing assets are additive, never a substitute for it.
  5. Business model & pricing hypothesis — how does money flow? Who pays, for what, how much, how often?
  6. Go-to-market motion — how do the first 10, 50, and 200 customers get acquired? Specific channels and mechanisms, not "content marketing". An established company can lean on existing channels, customer base, and brand here, which changes the first-customer motion.
  7. Market & competition — how big, who else plays here, what is the contrarian angle?
  8. Customer discovery status — how many real conversations, with whom, and what did you actually learn (not what you hoped to hear)?
  9. Risks, unknowns & falsification criteria — what would make you wrong? What would you have to see to kill this?
  10. Milestones & first experiments — what are you testing in the next 30/60/90 days, and what is the bigger 12-18 month milestone the funding is meant to reach (MVP in market, first paying enterprise clients, €X ARR)?
  11. The ask — the capital (or internal budget) needed to reach the Section 10 milestone, derived bottoms-up from the people, time, infra, and GTM spend it takes to get there. The number is an output of that build-up, not a guess. Plus the 1-2 hires that most change the trajectory, advisor and introduction asks, and design partners; or "none, bootstrapping because X". For a new line inside an existing company: budget, an executive sponsor, a ring-fenced team, mandate, build-vs-buy, and the internal hurdle rate the line must clear.
  12. Executive summary — written last, synthesised from the rest.

Weighting note. Pre-traction, Right to win (Section 4) is the single most load-bearing section. Everything else in the memo is a hypothesis; the team and its operator-market fit is the one thing that is already real, so it carries the most weight. If that section is thin or hand-wavy, the memo is weak no matter how polished the rest is. This holds for a corporate line too: the parent's assets do not rescue a line that lacks a credible team with real insight.

The ask is bottoms-up. When you reach Section 11, do not just record a number. Build it up from the Section 10 milestone: the roles and seniority needed, time to MVP and to first revenue, infra and tooling, and GTM hires and spend. Offer benchmark ranges (fully-loaded salaries by role, seniority, and region; typical infra and LLM costs) when the owner is unsure, compute a rough monthly burn, total, runway, and buffer, and tag it Hypothesis. Keep the full build-up — resourcing table and roll-up — in the memo's The Ask section: this is an investment memo, and the numbers are what let an investor evaluate the opportunity. Do not spin the budget out into the companion doc. See references/interview-questions.md (Section 11) and references/memo-template.md (Section 12).

Running research in parallel

For Domain Expert mode, research is mostly used to (a) validate or contradict claims the owner makes about competitors, market size, or "why now", and (b) surface anything the owner should know but might be missing. Be selective.

For Adjacent and Explorer modes, research is heavier. Run web searches for:

  • Direct and indirect competitors, their pricing, their positioning, any recent funding rounds.
  • Market size estimates from 2-3 independent sources (don't trust a single report).
  • "Why now" evidence: technology shifts, regulatory changes, public data on the problem's growth, recent shifts in customer behaviour.
  • Any obvious landmines: failed companies that tried this before, regulatory risks, incumbent responses.

Present research findings to the owner with context: "Here's what I found. Anything here change your view, or is this already priced in?"

See references/research-playbook.md for search strategies per section.

Capturing answers

Keep working notes as you go. A scratch structure like:

## Problem & target customer
- Owner said: [raw answer]
- Research found: [key facts, citations]
- Tag: [conviction / hypothesis]
- Open questions: [...]

This raw material feeds Phase 2.


Phase 2: Draft

Once the interview is complete, write the full memo in prose as a Word document using the docx skill (read its SKILL.md first). The argument lives in the writing; a table or a chart earns its place only by making real evidence land faster than a sentence would.

A visual amplifies, it does not substitute. One goes in only when the owner's real data supports it; never invent a number, a competitor, or a market curve to fill a slot. A missing chart is honest; a fabricated one corrupts the thesis, and the critical review exists partly to catch exactly that. Every visual carries its source. At most one per major section, never one per point.

Length target: 5-8 pages. Tight is better than long.

Tone: clear, honest, specific. First-person plural ("we believe") for conviction claims, hypothesis framing ("we hypothesise that…") for things not yet validated. Avoid adjectives that do no work ("revolutionary", "best-in-class"). Every paragraph should either describe reality or declare a bet.

Structure — follow references/memo-template.md exactly. Headings, subheadings, and one "Hypothesis" vs. "Conviction" tag per major claim. Short opening executive summary. Short closing "The ask" section.

Styling:

  • Professional, minimal. Black text, single accent colour.
  • Accent: 0F172A (near-black) for headings, 2563EB (blue) sparingly for emphasis, callout boxes, and the Conviction/Hypothesis tags.
  • Body text: 11pt, line spacing 1.15.
  • Headings: bold, 14pt for H1, 12pt for H2.
  • A small cover block at the top: company name, owner name, date, and a one-line positioning statement.
  • Tables: a header row in the accent, a light grid, no heavy fills.
  • No pitch-deck imagery. This is a document, not a visual artefact.

Prose style conventions (important):

  • No em dashes (—). Use commas, colons, semicolons, or periods instead. Em dashes are visual noise and a reliable tell for machine-generated prose. Hyphens (-) and en dashes (–) are fine for number ranges.
  • Smart quotes on (" " ' '), not straight quotes.
  • Active voice. First-person plural for conviction, hypothesis framing for untested claims.
  • No empty adjectives (no "revolutionary", "best-in-class", "world-class"). Every adjective describes reality or declares a bet.
  • Cite sources by name + year (e.g. "MIT NANDA 2025") when referencing studies or stats.
  • See references/memo-template.md for the full style guide.

Encoding note (docx-js): never use HTML entities in JavaScript string literals — write actual Unicode characters directly (e.g., ', ", ", —, –), or use Unicode escapes like \u2019. HTML entity syntax only works inside XML/HTML, not in JS strings.

Save the first draft as {company_name}_investment_memo_draft_{YYYY-MM-DD}.docx in the workspace. Then render it and look at every page before sharing (soffice --convert-to pdf, then pdftoppm to images): check that no table column is clipped, no heading is stranded at the foot of a page, and — most important — every number in a table matches the prose. A memo you have not looked at is not finished. Share the path with the owner and tell them to skim it before Phase 3.


Phase 3: Critical review

This is where the skill earns its keep. Once the draft exists, run it against the full critical review checklist in references/critical-review-checklist.md. Summarising the main pressure tests:

  • Fluff scan. Any claim without a name, number, or mechanism. Flag all.
  • Contradiction scan. Does section X assume Y while section Z assumes not-Y?
  • Wedge sharpness. Can the first product be described in one sentence? Is it narrow enough to build in a quarter?
  • Why-now test. If the "why now" was also true three years ago, it isn't one. What specifically changed recently?
  • Competitive honesty. Is the competitive landscape dismissed too quickly? Did the owner steelman the best incumbent response?
  • Right to win. Is the owner (or the team) a plausible winner of this market, or just a plausible builder? Pre-traction this is the most load-bearing section; flag it when assets or distribution are used to dodge whether this specific team has genuine operator-market fit.
  • GTM mechanism. Is there a concrete first-customer motion, or just channel-level handwaving?
  • Pricing sanity. Does the pricing imply a deal size that matches the GTM motion? (E.g., $500 ACV with an enterprise sales motion is a broken model.)
  • Market-size sanity. Is TAM calculated bottoms-up from real numbers, or top-down from a McKinsey chart?
  • Falsifiability. For each major belief, what would prove it wrong? If nothing would, it is a creed, not a belief.
  • Distribution of attention. Is the memo 70% product and 10% GTM? That's a red flag, and a thin Right to win section is a bigger one pre-traction.

After running the checklist, present your critique to the owner as a structured list:

  • Weak claims (with the specific quote and why it's weak).
  • Contradictions (with both quotes).
  • Missing evidence (what would make this section credible).
  • Questions you should sit with (2-3 hardest ones).

Do not rewrite the memo yet. Wait for the owner's responses.


Phase 4: Iterate

Go through the critique with the owner, one item at a time. For each:

  • Either the owner resolves it on the spot (they had an answer, it just wasn't in the draft), in which case update the memo.
  • Or it becomes an open question, in which case it flows to the companion document.

Update the memo in place. Keep a running changelog comment for your own use so you don't lose track of which sections were revised.

Run the critical review a second time after significant revisions. Stop when there are no major issues left, or when the owner says "good enough to ship" — whichever comes first. Two full review passes is usually the sweet spot.

The companion document

As critiques become open questions, write them into a separate Word document: {company_name}_open_questions_{YYYY-MM-DD}.docx. Structure:

  1. Open questions — grouped by memo section. Each with: the question, why it matters, and what kind of evidence would resolve it.
  2. Next experiments — 5-10 concrete things the owner could do in the next 30 days to close the biggest gaps. For each: the experiment, the hypothesis being tested, the success criterion, and a rough time cost. Ordered by priority (highest-information-per-hour first).
  3. Decisions log — brief notes on key decisions made during drafting (e.g., "We chose the SMB wedge over enterprise because of GTM velocity, revisit once we see retention data"). Future-them will thank present-them.

The full budget build-up belongs in the memo's The Ask section, not here. If specific cost assumptions are genuinely unresolved, capture them as open questions under item 1 — not as a separate budget section.

This companion doc is as important as the memo. The memo is the static snapshot; the companion doc is what drives the next 30 days of work.


Phase 5: Deliver

Save both files to the workspace root:

  • {company_name}_investment_memo_{YYYY-MM-DD}.docx
  • {company_name}_open_questions_{YYYY-MM-DD}.docx

Surface them with computer:// links.

In your final message:

  • Confirm both files are saved.
  • List the three sharpest questions the owner should go investigate this week, pulled from the companion doc.
  • Note which sections are still tagged as hypotheses rather than convictions, so the owner knows where the thesis is still load-bearing on assumption.
  • Offer the natural next step: "Want me to turn this into a pitch deck, a GTM plan, or a product strategy doc?"

Do not write an extensive postamble explaining the contents. The owner can read the document themselves.


Quality bar

The owner should be able to:

  1. Hand the memo to a smart stranger and have them understand the business thesis in 15 minutes.
  2. Hand the companion doc to themselves in three months and still know which bets are load-bearing on assumption.
  3. Reference the memo when writing a pitch deck, a GTM plan, or a hiring pitch — without having to re-derive the thinking.

If any of those three fail, go back and iterate.

Files in this skill

  • SKILL.md — this file.
  • references/memo-template.md — the full annotated section-by-section template used in Phase 2.
  • references/interview-questions.md — interview question bank, with mode-specific adjustments for Domain Expert / Adjacent / Explorer owners.
  • references/critical-review-checklist.md — full Phase 3 pressure-test checklist.
  • references/research-playbook.md — web-search strategies per memo section.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.