idea-validation-autopilot

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Idea Validation Autopilot

创意验证自动化工具（Idea Validation Autopilot）

Turn a rough idea into an evidence-backed build decision in one run.

只需一次运行，即可将初步想法转化为有证据支持的构建决策。

Overview

概述

This skill is a single orchestrator for:

idea clarification
market and competitor research
MVP scope definition
go/no-go style decision memo

Default behavior favors action over over-analysis:

ask as few questions as possible
run parallel research
output a build-ready decision packet

本Skill是一个集成协调器，可完成以下工作：

创意澄清
市场与竞品调研
MVP范围定义
启动/终止式决策备忘录

默认行为更倾向于行动而非过度分析：

尽可能少地提问
并行开展调研
输出可直接用于构建的决策包

When to Use

使用场景

Use this skill when:

user says they have many ideas but cannot decide efficiently
user wants "startup-like" process without paying for SaaS tools
user wants AI to drive research and synthesis, not just brainstorm
user asks for market validation + MVP boundaries + next execution steps

Do not use this skill when:

user already has validated requirements and only wants implementation planning
user wants only code generation with no discovery work

在以下场景中使用本Skill：

用户表示有多个想法但无法高效决策
用户希望采用“初创企业式”流程，且无需付费使用SaaS工具
用户希望AI主导调研与整合，而非仅进行头脑风暴
用户要求进行市场验证 + MVP边界定义 + 后续执行步骤

请勿在以下场景中使用本Skill：

用户已拥有经过验证的需求，仅需要实施规划
用户仅需要代码生成，无需探索性工作

Operating Defaults

操作默认值

If user context is missing, proceed with defaults instead of blocking:

Goal priority:
```
speed-to-learning > polish
```
Budget assumption:
```
near-zero external spend
```
Team assumption:
```
solo builder or very small team
```
Timebox assumption:
```
one focused discovery cycle
```

Only ask questions when missing data would invalidate the result (for example: unclear target user or regulated domain).

如果缺少用户上下文，将按默认值执行，而非停滞：

目标优先级：
```
speed-to-learning > polish
```
（学习速度优先于完善度）
预算假设：
```
near-zero external spend
```
（近乎零外部支出）
团队假设：
```
solo builder or very small team
```
（独立开发者或极小团队）
时间框假设：
```
one focused discovery cycle
```
（一个专注的探索周期）

仅当缺失数据会导致结果无效时才提问（例如：目标用户不明确或涉及受监管领域）。

Workflow

工作流程

Copy this checklist and track progress:

Progress
- [ ] Step 1: Normalize idea into problem hypothesis
- [ ] Step 2: Run 4 parallel research tracks
- [ ] Step 3: Grade evidence quality and resolve contradictions
- [ ] Step 4: Produce decision scorecard and verdict
- [ ] Step 5: Define MVP scope and exclusions
- [ ] Step 6: Define first experiments and stop rules
- [ ] Step 7: Deliver final report using template

复制以下清单并跟踪进度：

Progress
- [ ] Step 1: Normalize idea into problem hypothesis
- [ ] Step 2: Run 4 parallel research tracks
- [ ] Step 3: Grade evidence quality and resolve contradictions
- [ ] Step 4: Produce decision scorecard and verdict
- [ ] Step 5: Define MVP scope and exclusions
- [ ] Step 6: Define first experiments and stop rules
- [ ] Step 7: Deliver final report using template

Step 1: Normalize the idea

步骤1：标准化创意

Convert raw idea into this structure:

target user
painful job-to-be-done
current workaround
why-now trigger
value promise in one sentence

If unclear, propose your best assumption and mark it explicitly.

将原始创意转化为以下结构：

目标用户
亟待解决的核心任务
当前的替代方案
即时启动的触发因素
一句话价值主张

如果信息不明确，提出你的最佳假设并明确标记。

Step 2: Run 4 parallel research tracks

步骤2：并行开展4项调研任务

Dispatch four independent subagents (or equivalent parallel workers).

User/Problem Research

Find who feels the pain and how urgently.
Capture behavioral evidence, not just opinions.

Market/Competitor Research

Map direct/adjacent alternatives, pricing, positioning, switching cost.
Identify market gap with realistic differentiation.

Business Model/Risk Research

Estimate willingness-to-pay signals, acquisition path, and major risks.
Flag legal/compliance/data-access blockers early.

MVP/Technical Feasibility Research

Define thinnest viable product delivering the core job.
Identify build constraints, integration risks, and timeline risk.

调度四个独立的子Agent（或等效的并行工作单元）：

用户/问题调研

找出受困扰的人群及困扰的紧迫程度
收集行为证据，而非仅依赖观点

市场/竞品调研

梳理直接/间接替代方案、定价、定位、转换成本
识别具有现实差异化的市场空白

商业模式/风险调研

估算付费意愿信号、获客路径及主要风险
尽早标记法律/合规/数据访问方面的障碍

MVP/技术可行性调研

定义可交付核心任务的最简可行产品（MVP）
识别构建约束、集成风险及时间线风险

Step 3: Grade evidence quality

步骤3：评估证据质量

Use evidence tiers:

```
Tier A
```
: behavioral or monetary signal (payment, waitlist intent with commitment, repeated real usage)
```
Tier B
```
: strong secondary evidence (credible reports, robust competitor/user data)
```
Tier C
```
: weak signal (opinions, generic trend articles, unsupported claims)

Rules:

critical claims need at least two independent sources
if evidence is weak, lower confidence regardless of narrative quality

使用以下证据层级：

```
Tier A
```
：行为或货币信号（付款、带有承诺的等待列表意向、重复实际使用）
```
Tier B
```
：可靠的间接证据（可信报告、详实的竞品/用户数据）
```
Tier C
```
：弱信号（观点、通用趋势文章、无支撑的主张）

规则：

关键主张需要至少两个独立来源
如果证据薄弱，无论叙述质量如何，均降低置信度

Step 4: Score and decide

步骤4：评分与决策

Score 0-100 using weighted dimensions:

Dimension	Weight
Problem severity and frequency	25
Distribution reachability	20
Willingness-to-pay potential	20
MVP speed/feasibility	20
Strategic differentiation	15

Scoring rules (fixed):

each dimension score is
```
0..100
```
```
weighted_i = score_i * weight_i / 100
```
```
total_score = round(sum(weighted_i), 1)
```
map verdict from
```
total_score
```
using the bands below

Verdict bands:

```
80-100
```
: Build now
```
60-79
```
: Validate-first (run targeted tests before building)
```
40-59
```
: Pivot
```
<40
```
: Drop

使用加权维度进行0-100分评分：

维度	权重
问题的严重性与发生频率	25
触达用户的可能性	20
付费意愿潜力	20
MVP开发速度/可行性	20
战略差异化	15

评分规则（固定）：

每个维度的分数为
```
0..100
```
```
weighted_i = score_i * weight_i / 100
```
```
total_score = round(sum(weighted_i), 1)
```
根据以下分数区间映射最终结论

结论区间：

```
80-100
```
：立即构建
```
60-79
```
：先验证（在构建前开展针对性测试）
```
40-59
```
：调整方向
```
<40
```
：放弃

Step 5: Define MVP scope

步骤5：定义MVP范围

Use strict scope slicing:

```
Must
```
: smallest set proving core value
```
Should
```
: useful but deferrable
```
Won't (now)
```
: explicitly excluded features

Output a 2-week implementation target:

week 1: build core flow
week 2: launch to first users and collect signals

使用严格的范围划分：

```
Must
```
：证明核心价值的最小功能集合
```
Should
```
：有用但可延迟的功能
```
Won't (now)
```
：明确排除的功能

输出2周实施目标：

第1周：构建核心流程
第2周：向首批用户发布并收集反馈信号

Step 6: Define experiments and stop rules

步骤6：定义实验与终止规则

For top risks, define:

experiment
pass threshold
fail threshold
next action if pass/fail

Keep experiments cheap and fast. Favor reversible steps.

针对最高风险，定义：

实验内容
通过阈值
失败阈值
通过/失败后的下一步行动

保持实验低成本、快节奏。优先选择可逆步骤。

Step 7: Deliver final report

步骤7：交付最终报告

Use

assets/final-report-template.md

Output path rules:

if
```
reports/
```
does not exist, create it first (
```
mkdir -p reports
```
)

write report to

reports/YYYY-MM-DD-<idea-slug>-idea-validation.md

Required output qualities:

explicit assumptions table
explicit unknowns
citations and dated evidence
final recommendation plus next 7-day action plan

使用

assets/final-report-template.md

模板。

输出路径规则：

如果
```
reports/
```
目录不存在，先创建（
```
mkdir -p reports
```
）

将报告写入

reports/YYYY-MM-DD-<idea-slug>-idea-validation.md

报告必备要素：

明确的假设表
明确的未知事项
引用及带日期的证据
最终建议及未来7天行动计划

Common Failure Modes

常见失败模式

Over-research without decisions

Fix: enforce scorecard and verdict section every run.

Generic competitor list with no switching analysis

Fix: include why users switch or stay.

MVP too large

Fix: require "what can be deleted" before finalizing scope.

False confidence from weak sources

Fix: downgrade to Tier C and force validation-first verdict.

过度调研却不做决策

解决方法：每次运行都强制生成评分卡与结论部分

竞品列表泛泛而谈，未分析用户转换原因

解决方法：加入用户转换或留存原因的分析

MVP范围过大

解决方法：在最终确定范围前，必须明确“可删除的内容”

因来源薄弱而产生虚假信心

解决方法：将其降级为Tier C，并强制给出“先验证”的结论

Quick Command Patterns

快速命令模式

Adapt to available tools:

web search + fetch for sources
repository/API lookup for existing solutions
parallel subagents for independent tracks
markdown report output in project
```
reports/
```

If one tool is unavailable, continue with the best fallback and document the limitation in assumptions.

适配可用工具：

网页搜索 + 抓取来源信息
查找现有解决方案的代码库/API
使用并行子Agent开展独立调研任务
在项目
```
reports/
```
目录中输出Markdown报告

如果某一工具不可用，使用最佳替代方案继续，并在假设部分记录该限制。