skill

matteocervelli

更新日 Yesterday

65 閲覧

メタgeneral

について

このClaudeスキルは、製品評価の回答を構造化されたJSON形式で段階的に収集し検証します。新製品アイデアの評価、Go/No-Go判断の実施、チームからの構造化された入力収集を目的として設計されています。本スキルは4つの主要セクション（WHY、WHO、WHAT、GO/NO-GO）で質問を体系化し、検証済みのJSON回答を段階的に構築していきます。

クイックインストール

Claude Code

推奨

プラグインコマンド推奨

/plugin add https://github.com/matteocervelli/llms

Git クローン代替

git clone https://github.com/matteocervelli/llms.git ~/.claude/skills/skill

このコマンドをClaude Codeにコピー＆ペーストしてスキルをインストールします

ドキュメント

Answer Collector Skill

Purpose: Incrementally collect and validate product assessment responses in structured JSON format.

When to Use

Evaluating new product ideas with rigorous criteria
Conducting go/no-go assessments before committing resources
Building a decision audit trail for product decisions
Gathering structured input from teams or stakeholders
Progressive refinement of product hypotheses

How It Works

1. Reading Questions

Questions are organized in 4 sections (questions.md):

WHY (4 Q's): Problem, strategy, resources, timing
WHO (4 Q's): User, access, economics, scale
WHAT (5 Q's): Outcome, monetization, success metrics, fit, risk
GO/NO-GO (4 criteria): Checklist for final decision

Each question is numbered 1-17.

2. Writing JSON Incrementally

Start with a template and add answers one at a time:

{
  "metadata": {
    "product_name": "Your Product Name",
    "created_at": "2025-11-03T00:00:00Z",
    "status": "in_progress"
  },
  "answers": {
    "why_section": {
      "q1_problem_evidence": "Answer here..."
    }
  }
}

Build incrementally:

Add one answer per interaction
Preserve all previous answers
Update last_updated timestamp
Track completion_percentage in metadata

3. Validation Logic

Auto-calculate:

answered_questions: Count non-empty answers
completion_percentage: (answered_questions / 17) × 100
go_no_go_result: "go" if all 4 checklist items true, else "no_go" or "pending"

Validation rules:

All text answers must be non-empty and substantive
Checklist items (q14-q17) must be boolean (true/false)
Metadata fields (product_name) required to start
All timestamps in ISO 8601 format

Quick Reference

Section	Questions	Type
WHY	1-4	Text
WHO	5-8	Text
WHAT	9-13	Text
GO/NO-GO	14-17	Boolean

Usage Pattern

Initialize: Create JSON with metadata and product_name
Collect: Answer one question, validate, save
Review: Check completion_percentage and go_no_go_result
Decide: When all answers complete, review go_no_go_result

File Structure

~/.claude/skills/answer-collector/
├── SKILL.md           # This file
├── questions.md       # The 17 assessment questions
├── schema.json        # JSON validation schema
└── assessments/       # (optional) Stored assessment JSONs
    └── product-name.json

GitHub リポジトリ

matteocervelli/llms

パス: .claude/skills/answer-collector