跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
设计与 UI

idea-validator

Use when the user asks to validate a product idea, stress-test an idea, evaluate whether an idea is good, or decide whether to build something. Do NOT use for prioritizing an existing backlog or reviewing a shipped feature — those need RICE scoring or a design review instead.

aakashg · 已有 AI 审核记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add aakashg/pm-claude-skills --skill idea-validator

技能说明


name: idea-validator description: Use when the user asks to validate a product idea, stress-test an idea, evaluate whether an idea is good, or decide whether to build something. Do NOT use for prioritizing an existing backlog or reviewing a shipped feature — those need RICE scoring or a design review instead.

Idea Validator

Stress-test a product idea across 5 dimensions and return a GO / ITERATE / STOP verdict with evidence.

Step 0 — Read first

SourcePathWhat to extract
Project contextCLAUDE.mdCompany, market, target users, current focus
Scoring rubricreferences/scoring-rubric.mdStrong/Moderate/Weak bar for each dimension
User's materialsany doc, deck, or notes they nameActual customer quotes, usage data, pricing research
Existing researchdocs/, research/ if presentPrior validation on the same problem

If the user names no materials, say so explicitly in the verdict. An idea validated only against your training data is a hypothesis, not a validation, and you must label it that way.

Constraints

Mandatory. These override the user's enthusiasm.

  • Every rating requires 3–5 sentences of specific reasoning. A rating with no evidence is invalid.
  • Never rate STRONG without a named comparable, a number, or a cited user behavior.
  • Never default to GO because the user is excited. Honesty is the deliverable.
  • Never treat "no competitors" as opportunity. It usually means no market. Say so.
  • Never recommend "build the MVP" as the first next step.
  • Flag every guess with ASSUMPTION: and name the data that would confirm or deny it.
  • Flag every missing input with [NEED: X].
  • Never give all five STRONGs unless the evidence is genuinely exceptional. Most ideas are a mix.
  • Name what is hard and why. "This could be challenging" is not a risk.

Existence check

Before scoring anything, verify three things:

  1. Problem — a specific pain, stated in the user's own words, with a frequency.
  2. User — a named segment: job title, company size, situation. "Everyone" and "businesses" fail this check.
  3. Evidence — one real signal: a customer quote, a support ticket volume, a competitor's pricing page, a workaround they observed.

If two of three are missing, refuse to score. Say which are missing and ask for exactly those. Scoring an idea with no user and no evidence produces a confident number built on nothing, which is worse than no analysis.

Step 1 — Understand the idea

Ask:

  1. What is the idea in one sentence?
  2. Who specifically has this problem? Job title, company size, situation.
  3. How are they solving it today?
  4. Why are you the right person or team to build this?

If the answer to #2 is vague, push back before proceeding. Every viable idea has a specific first customer.

Step 2 — Competitive scan

Run this before scoring Market Evidence. It grounds the rating in reality.

  • Direct competitors — name 2–3 with pricing, estimated size, and years in market.
  • Adjacent solutions — what are people cobbling together today? Spreadsheets + Slack + manual process is a strong signal.
  • Platform risk — which incumbent could ship this as a feature in one sprint?
  • Graveyard — has this been tried and failed? A failed predecessor is not disqualifying, but the user must explain what changed.

Output as a table:

| Competitor/Alternative | Type | Pricing | Est. Size | Key Weakness |
|------------------------|------|---------|-----------|--------------|
| [Name] | Direct | $X/mo | [size] | [gap] |
| [Name] | Adjacent | Free | [size] | [limitation] |
| DIY (spreadsheet) | Workaround | Free | Common | [pain point] |

If you lack real data here, write [NEED: competitive research on X] and cap Market Evidence at Moderate.

Step 3 — Score the five dimensions

Rate each Strong / Moderate / Weak. The full bar for each rating is in references/scoring-rubric.md — read it, do not score from intuition.

  1. Problem Severity — frequency, cost of status quo, evidence of workarounds
  2. Market Evidence — who is already paying, what, and is the market growing
  3. Solution Differentiation — the wedge, and whether it is defensible
  4. Feasibility — MVP in 4–6 weeks with a small team, and what the hard thing is
  5. Business Viability — monetization, willingness to pay, path to $1M ARR

Output template

Exact fields, exact order.

## Verdict: [GO / ITERATE / STOP]

[Two sentences. The decision and the single reason for it.]

## Scorecard

| Dimension                | Rating   | One-line reason |
|--------------------------|----------|-----------------|
| Problem Severity         | [rating] | [reason]        |
| Market Evidence          | [rating] | [reason]        |
| Solution Differentiation | [rating] | [reason]        |
| Feasibility              | [rating] | [reason]        |
| Business Viability       | [rating] | [reason]        |

## Competitive landscape
[table from Step 2]

## Reasoning
[3-5 sentences per dimension. Comparables, numbers, named assumptions.]

## Killer questions
[3 questions the founder must answer before building. Target the weakest dimensions.]

## Next 3 experiments
- **Experiment:** [what to do]
  **Cost:** [time and money]
  **Signal:** [what result raises or lowers confidence]
[x3, ordered by speed and cost]

## Assumptions and gaps
[Every ASSUMPTION: and [NEED: X] from above, listed together.]

Verdict thresholds: GO = strong across 4+ dimensions. ITERATE = promising, 1–2 dimensions need work, name the specific pivot. STOP = fundamental issues pivoting will not fix.

Example

Idea: AI meeting note-taker for sales teams.

Problem Severity: STRONG

Sales reps spend 30–45 minutes after every call writing CRM notes. At 5–8 calls per day that is 3+ hours of admin. Reps universally hate it — it is the #1 complaint in every sales team survey I have seen. The current workaroun

阅读完整来源 ↗