Skip to content
Back to skills

Validate Business Model

ASecurity

Validate a Sui product business model (who pays, how much, why). Use when the user wants to validate monetization, pricing, or business model.

  • 12 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added May 27, 2026
developmentbash

Works with

  • claude code
  • cursor
  • cli

Security analysis

A100/100

Pro scans all 4 files and shows the line behind each finding

Scanned May 27, 2026

npx -y skills add pivyme/suiperpower --skill validate-business-model --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Validate Business Model?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Validate Business Model
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/pivyme-validate-business-model/badge)](https://www.skillsdirectory.com/skills/pivyme-validate-business-model)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: validate-business-model
description: Validate a Sui product business model (who pays, how much, why). Use when the user wants to validate monetization, pricing, or business model.
---

## Preamble (run first)

```bash
# Suiperpower telemetry. Routes through the CLI so projects.json + Convex stay in sync.
# Silent on failure so the skill never blocks. Opt out: SUIPERPOWER_TELEMETRY=off.
#
# AGENT NOTE: when this skill finishes, run the matching completion command:
#   suiperpower track validate-business-model build completed
# Or use "failed" / "aborted" if it ended that way.
command -v suiperpower >/dev/null 2>&1 && suiperpower track validate-business-model build started >/dev/null 2>&1 &
true
```

If `TEL_PROMPTED` is `no`, before doing real work, ask the user:

> Help suiperpower get better. We track which skills get used and how long they take. No code, no file paths, no PII. Change anytime in `~/.suiperpower/config.json`.
>
> A) Sure, anonymous
> B) No thanks

Write the answer to `~/.suiperpower/config.json` `telemetryTier` field and create `~/.suiperpower/.telemetry-prompted`. Then continue.

## What this skill does

Walks the user through five concrete questions about how the project makes money. Refuses to write a "validated business model" output if any answer is hand-wavy or speculative. The point is to force the user to either commit to a real answer or accept that they have not yet validated this dimension. Both outcomes are useful; pretending to validate when nothing is validated is the slop.

`deploy-to-mainnet` reads this skill's output and refuses to deploy if no business-model output exists.

## When to use it

- Pre-mainnet, when the project is about to be exposed to real users.
- During pitch prep when the user is about to claim a business model in front of investors or judges.
- Before applying for grants or hackathon prizes.

## When NOT to use it

- Pre-MVP, when the product is not built. Defer until there is a working v1.
- For research-only projects that the user has explicitly framed as non-commercial.
- For pre-validated existing products being ported to Sui (the model already exists; document it instead).

If you activated this and the user actually wants something else, consult `skills/SKILL_ROUTER.md` and hand off.

## Inputs

- `.suiperpower/idea-context.md` for context.
- The current product state (early build, MVP, near-launch).
- Whether the user has talked to any prospective payers.

## Outputs

A `.suiperpower/business-model.md` with explicit answers to all five questions, or an explicit `unvalidated` marker on each unanswered question. Format:

```markdown
## Business model, <timestamp>

### Q1, who pays?
- answer: <user | developer | partner | treasury | advertiser | unvalidated>
- evidence: <one sentence with citation>
- confidence: <high | medium | low>

### Q2, how much?
- pricing model: <per-tx | per-month | per-feature | per-volume | one-time | unvalidated>
- price point: <number with currency>
- evidence: <one sentence>
- confidence: <high | medium | low>

### Q3, why would they pay you instead of an alternative?
- alternative: <named competitor or status quo>
- our advantage: <one sentence>
- evidence: <one sentence>
- confidence: <high | medium | low>

### Q4, what is the unit economic at one paying user?
- revenue per user per month: <number>
- variable cost per user per month: <number including Walrus / DeepBook / RPC / hosting>
- gross margin: <percent>
- confidence: <high | medium | low>

### Q5, what test could falsify this in the next two weeks?
- test: <one sentence>
- success threshold: <number, e.g. "5 paying users at $X">
- planned date: <yyyy-mm-dd>

### Verdict
- validated: <yes | partial | no>
- if partial or no: which questions are open
```

## Workflow

1. **Read idea-context.md**
   - Pull the chosen idea, target user, sponsor track if any.

2. **Ask Q1**
   - "Who specifically pays you? Pick one of: user, developer, partner, treasury, advertiser. If multiple, pick the largest."
   - Reject "everyone" and "depends".

3. **Ask Q2**
   - "Concretely, how much? Per transaction in basis points? Per month subscription? One-time?"
   - Reject ranges that span more than 2x.

4. **Ask Q3**
   - "Why would the payer pay you instead of <named alternative>?"
   - Reject "we are better" and "we are cheaper" without specifics.

5. **Ask Q4**
   - "If you had ONE paying user at the price in Q2, what is your monthly margin?"
   - Force the math. Variable costs include any sponsor protocol fees the product passes through.

6. **Ask Q5**
   - "What experiment, runnable in two weeks, would tell you if Q1-Q3 are right?"
   - Reject "build the product and see". The test must be cheaper than building.

7. **Score and write**
   - For each question, mark high / medium / low confidence.
   - For each question with no concrete answer, mark `unvalidated`.
   - Write the verdict honestly. Two or more `unvalidated` = `no`. One = `partial`. Zero = `yes`.

8. **Hand off**
   - If `validated: yes`, the user is cleared for `deploy-to-mainnet` (combined with the other gates).
   - If `partial` or `no`, recommend `will-real-users-pay` to run the cheap experiment.

## Quality gate (anti-slop)

Before reporting done:

- Did every question get a concrete answer or an explicit `unvalidated`? (Hand-wavy answers count as `unvalidated`.)
- Is Q4 backed by actual math, not "should be profitable"?
- Is Q5 a test that costs less than building the product, with a date and a threshold?
- Did the writeback happen?
- If the verdict is `yes`, can the skill point at the evidence in each question?

If any answer is no, the skill keeps working.

## References

On-demand references (load when relevant to the user's question):

- `references/payer-types.md`: Definitions and examples of each payer type.
- `references/cost-checklist.md`: Variable costs to include in Q4 (sponsor protocol fees, RPC, hosting, support).

## Use in your agent

- Claude Code: `claude "/suiper:validate-business-model <your message>"`
- Codex: `codex "/validate-business-model <your message>"`
- Cursor: paste a chat message that includes a phrase like "validate my business model", or load `~/.cursor/rules/validate-business-model.mdc` and reference it.

If you activated this and the user actually wants something else, consult `skills/SKILL_ROUTER.md` and hand off.

Files in this skill

  • SKILL.md6.2 KB
  • agents/openai.yaml316 B
  • references/cost-checklist.md2.8 KB
  • references/payer-types.md3.4 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.