agentleFS
YehudaFrankelYehudaFrankel/clankbrain/.claude/skills/product-risk/SKILL.md

product-risk

Skill16 starsChanged 7 months agoMITView on GitHub

Four Big Risks product validation. Two modes: evaluate (score an existing product Red/Yellow/Green) or create (validate idea + generate working prototypes + full project scaffolding). Triggers only when asked: /product-risk, evaluate this product, validate this idea, is this worth building, product risk check, run the four risks, score this product, rate this idea. Does not start on everyday phrases like new product or what are the risks.

The file

Show its rules and text

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈17

    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:

    If the user says "evaluate", "score", "check", "audit", or names an existing project → Evaluate Mode If the user says "create", "new", …

    line 15
  2. Distinctive:

    If the user says "create", "new", "build", "idea", "I want to make" → Create Mode If ambiguous, ask: "Are we evaluating an existing product …

    line 16
  3. Distinctive:

    If ambiguous, ask: "Are we evaluating an existing product or validating a new idea?"

    line 17
  4. Distinctive:

    Ask: "Which product are we evaluating?"

    line 25
  5. Distinctive:

    If it's a project in this workspace, read its plan file, STATUS.md, or CLAUDE.md for context.

    line 27
  6. Distinctive:

    Ask these questions one at a time.

    line 90
  7. Distinctive:

    Run the same 4-risk evaluation from Evaluate Mode, but against the idea — not an existing product.

    line 101
  8. Distinctive:

    Kill gate: If 2+ risks are red, stop and say:

    line 105
  9. Distinctive:

    Do NOT proceed to Phase 3 with 2+ red risks.

    line 108
  10. Distinctive:

    If 0-1 risks are red, proceed.

    line 110
  11. Distinctive:

    If the user can't name 10 in under a minute, distribution risk is real.

    line 128
  12. Distinctive:

    Write the answer as a 2-week scope statement.

    line 135
  13. Distinctive:

    If any of the 3 is "I don't know" or handwavy, flag as Yellow and note what to resolve before building.

    line 148
  14. Distinctive:

    Wait for approval before generating.

    line 163
  15. Very specific:

    Add <!-- PROTOTYPE — [Product Name] — generated by /product-risk create --> at the top

    line 176
  16. Distinctive:

    Use the design direction from Phase 3

    line 177
  17. Distinctive:

    Every interactive element should work (buttons open modals, tabs switch, forms validate)

    line 178
  18. Very specific:

    Ask the user where to place the prototype files.

    line 181
  19. Distinctive:

    If the user says "skip the questions" or "just build it" — push back once: "The questions take 10 minutes.

    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

Skill: product-risk

Trigger: /product-risk, "evaluate this product", "validate this idea", "new product", "is this worth building", "product risk check", "run the four risks"


Detect Mode

If the user says "evaluate", "score", "check", "audit", or names an existing project → Evaluate Mode If the user says "create", "new", "build", "idea", "I want to make" → Create Mode If ambiguous, ask: "Are we evaluating an existing product or validating a new idea?"


EVALUATE MODE — Score an existing product

Step 1: Identify the product

Ask: "Which product are we evaluating?"

If it's a project in this workspace, read its plan file, STATUS.md, or CLAUDE.md for context. If it's external, ask the user to describe it in 2-3 sentences.

Step 2: Walk through the Four Risks

For each risk, ask 2-3 targeted questions. Listen to the answers. Don't lecture — extract signal.

1. Value Risk (Will they buy it?)

  • Who specifically is the target user? Not "developers" — which developers, doing what, hitting what pain?
  • What do they do today without this product? (If the answer is "nothing" or "it's fine" — red flag)
  • Would they pay for it? How much? (If the answer is vague — yellow flag)

2. Usability Risk (Can they use it?)

  • How many steps from "I found this" to "I got value"? (More than 5 = yellow. More than 10 = red)
  • Does it require knowledge the target user doesn't have?
  • Has anyone outside the builder actually tried it?

3. Feasibility Risk (Can we build it?)

  • What's the core technical challenge? Is it solved or speculative?
  • How long to MVP? (If the answer is "6 months" for a solo builder — red flag)
  • What dependencies exist? (Third-party APIs, platforms, approvals)

4. Viability Risk (Should we do this?)

  • How does it make money? When?
  • What's the competitive landscape? Why would someone choose this over alternatives?
  • Does the builder have an unfair advantage? (Domain expertise, existing audience, distribution)

Step 3: Score and output

Rate each risk:

  • Green — validated or low risk
  • Yellow — plausible but unproven, needs testing
  • Red — serious concern, could kill the product

Output format:

## Product Risk Scorecard — [Product Name]
Date: [YYYY-MM-DD]

| Risk | Score | Summary |
|------|-------|---------|
| Value (Will they buy?) | [color] | [one line] |
| Usability (Can they use?) | [color] | [one line] |
| Feasibility (Can we build?) | [color] | [one line] |
| Viability (Should we do this?) | [color] | [one line] |

### Biggest risk: [which one]
[2-3 sentences on what specifically is unproven and how to test it]

### Next action
[One concrete thing to do this week to de-risk the biggest gap]

Save the scorecard to the memory directory as product-risk-[name].md if the user wants to keep it.

Stop here for evaluate mode.


CREATE MODE — Validate + Prototype + Scaffold

Phase 1: The Idea (5 minutes)

Ask these questions one at a time. Wait for each answer before proceeding.

  1. "What problem are you solving? Describe the pain, not the solution."
  2. "Who has this problem? Be specific — job title, situation, frequency."
  3. "What do they do today? What's the current workaround?"
  4. "Why would they switch to your thing?"

Summarize back: "So you're building [X] for [Y] because [Z]. Right?"

Phase 2: Four Risk Gate (10 minutes)

Run the same 4-risk evaluation from Evaluate Mode, but against the idea — not an existing product. Be honest. If a risk is red, say so plainly:

"Value risk is red — you described the user as 'everyone who codes.' That's not a customer, that's a hope. Who specifically would pay $X/month for this? Give me a name or a job title."

Kill gate: If 2+ risks are red, stop and say:

"Two risks are red. Building a prototype now would waste time. Here's what to validate first: [specific actions]. Come back when you have answers."

Do NOT proceed to Phase 3 with 2+ red risks. The skill's job is to prevent building things nobody wants.

If 0-1 risks are red, proceed.

Phase 2b: Go-To-Market Reality (5 minutes)

The Four Risks pass judges whether to build. This pass judges how you'd actually launch — three questions that kill more products than any of the Four Risks. Skip only if the user is shipping to an existing audience you've already validated.

a) Real Competition — who's the actual enemy?

  • "What do people do TODAY to solve this? Include the boring answers (spreadsheet, doing nothing, WhatsApp group, paying someone)."
  • "Name direct competitors (same category) AND indirect (different category, same job-to-be-done)."
  • "Which one is the real enemy — the competitor users will actually compare you to?"

The status-quo ("they do nothing / they live with it") counts as a competitor. Often the hardest one.

b) First 10 Customers — where are they, how do you reach them?

  • "Name 10 specific people (or narrow segments) who would try this in the first month. Not 'small business owners' — 10 names or one tight group you can list by Saturday."
  • "Where do those 10 hang out online or offline?"
  • "What's the manual outreach plan — DMs, email, showing up? (Not 'we'll run ads' — ads are for later.)"

If the user can't name 10 in under a minute, distribution risk is real. Flag as Yellow even if all Four Risks were Green.

c) 2-Week MVP — what's the minimum that proves the idea?

  • "If you had to ship something a real user could try in 2 weeks, what's the ONE thing it does?"
  • "What can you cut? What CAN'T you cut?"
  • "What's the launch plan for day 1 — not marketing, just 'how does the first user get in'?"

Write the answer as a 2-week scope statement. Anything that doesn't fit goes on a "later" list.

Output of Phase 2b — append to scorecard:

### Go-To-Market Reality
| Question | Answer |
|----------|--------|
| Real enemy | [the thing to beat] |
| First 10 plan | [who + where + how to reach] |
| 2-week MVP | [single-sentence scope] |

If any of the 3 is "I don't know" or handwavy, flag as Yellow and note what to resolve before building.

Phase 3: Product Shape (10 minutes)

Now that the idea is validated, define what to build:

  1. "What are the 3-5 core screens a user would see?"
  2. "What's the one action that delivers the core value? (The thing they came for)"
  3. "What's the simplest version that proves the idea works?"

From the answers, propose:

  • Page list (e.g., Landing, Dashboard, Settings)
  • Core user flow (e.g., Sign up -> Connect repo -> See first sync)
  • Design direction (ask: "Warm and friendly, or clean and technical? Dark or light? Any products you want it to feel like?")

Wait for approval before generating.

Phase 4: Generate Prototypes (20 minutes)

Build working HTML/CSS/JS prototypes. Not wireframes — real interactive pages with:

  • Actual navigation between pages
  • Realistic sample data (not lorem ipsum)
  • Responsive layout (mobile-first)
  • Click handlers, modals, transitions
  • Design tokens as CSS variables (easy to re-skin later)

Rules for prototypes:

  • One HTML file per page (self-contained — inline CSS and JS)
  • Add <!-- PROTOTYPE — [Product Name] — generated by /product-risk create --> at the top
  • Use the design direction from Phase 3
  • Every interactive element should work (buttons open modals, tabs switch, forms validate)
  • Include realistic content that shows what the product actually does

Ask the user where to place the prototype files. Default to the project root or a prototypes/ directory.

Phase 5: Project Scaffolding

Generate full project context so the next session starts with everything:

1. Plan file → save to the memory plans/ directory

# Plan: [Product Name]
**Status:** Draft
**Created:** [date]

## Problem
[From Phase 1 answers]

## Risk Assessment
[Scorecard from Phase 2]

## Core Screens
[From Phase 3]

## Technical Spec
[Stack recommendation based on what was discussed]

## Open Questions
[Any unresolved items from the conversation]

2. Decisions file entry → append to decisions.md in memory

## [date] - [Product Name]: initial product decisions
**Decision:** [Key choices made during the session]
**Rationale:** [Why, based on 4-risk analysis]

3. Memory index entry → append to MEMORY.md

- [Product Name plan](plans/[product-name].md) — Status: Draft. [one-line summary]

Phase 6: Report

## Product Risk Create — Complete

### Idea: [name]
[One sentence]

### Risk Scorecard
| Risk | Score |
|------|-------|
| Value | [color] |
| Usability | [color] |
| Feasibility | [color] |
| Viability | [color] |

### Generated
- [N] prototype pages: [list filenames]
- Plan file: [path]
- Decisions logged
- Memory index updated

### Next steps
1. [Most important thing to validate]
2. [Second priority]
3. [Third priority]

Notes

  • The kill gate in Phase 2 is the most important part of this skill. Building prototypes for a bad idea is worse than building nothing.
  • Prototypes are starting points, not finished products. They should look real enough to test with users but be understood as disposable.
  • If the user says "skip the questions" or "just build it" — push back once: "The questions take 10 minutes. Building the wrong thing takes 10 sessions. Which is more expensive?" If they insist, proceed but note the risks as yellow/unvalidated.

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.