NPD Validator — Claude Code Plugin

Multi-agent product validation pipeline that orchestrates 8 specialized subagents to evaluate new product concepts and produce evidence-based launch recommendations.

What It Does

When you run /npd-validator:validate-product, the plugin:

1. Research Coordinator gathers market data, competitor profiles, and category facts
2. 5 Independent Evaluators each run in their own context window — no anchoring bias:
   • Market Demand Analyst (category size, demand signals, timing)
   • Competitive Intelligence Analyst (white space, competitor weakness, barriers)
   • Brand Fit Strategist (brand alignment, portfolio coherence, audience fit)
   • Commercial Viability Analyst (pricing power, unit economics, channels)
   • Consumer & Trend Analyst (social sentiment, trend momentum, cultural fit)
3. Devil's Advocate reads all 5 evaluations and builds the strongest case NOT to launch
4. Consensus Director identifies conflicts, challenges evaluators, and produces the final report

Each evaluator is a real subagent with its own context window — not one model pretending to be multiple perspectives. This ensures genuine analytical independence.

Install

claude --plugin-dir /path/to/npd-validator-plugin

Or from a marketplace:

/plugin install npd-validator

Usage

/npd-validator:validate-product A caffeine-infused body moisturizer targeting 25-35 year old
women, priced at $28-35, positioned as clean beauty with an energy boost angle, for Frank Body

Or start with a rough idea:

/npd-validator:validate-product I think there's a gap in men's skincare for something simple

Output

A comprehensive validation report with:

• Verdict: GO / CONDITIONAL GO / REVISIT / NO-GO
• Composite Score: weighted average of 15 scored dimensions
• Master Scorecard: all 15 dimensions with scores and confidence
• Competitor Map: top competitors with positioning analysis
• Pricing Benchmarks: category pricing tiers
• Trend Evidence: social signals and cultural alignment
• Devil's Advocate Case: strongest counter-arguments with kill shot rating
• Deliberation Log: how conflicts were identified and resolved
• Recommended Next Steps: specific actions

Iteration

After the report, you can:

• Pivot: "What if we repositioned as premium?" → re-runs pipeline
• Deepen: "Dig deeper on competitive landscape" → re-runs that evaluator
• Challenge: "I disagree with the brand fit score because..." → evaluator re-assesses
• Reweight: "Brand fit matters more for us" → recalculates composite

Architecture

npd-validator-plugin/
├── .claude-plugin/
│   └── plugin.json               ← Plugin manifest
├── skills/
│   ├── validate-product/
│   │   └── SKILL.md              ← Main /validate-product command (orchestrator)
│   └── npd-methodology/
│       └── SKILL.md              ← Scoring rubrics and templates (reference)
├── agents/
│   ├── npd-research-coordinator.md
│   ├── npd-market-demand.md
│   ├── npd-competitive-intel.md
│   ├── npd-brand-fit.md
│   ├── npd-commercial-viability.md
│   ├── npd-consumer-trends.md
│   ├── npd-devils-advocate.md
│   └── npd-consensus-director.md
└── README.md

Requirements

• Claude Code with subagent support
• Web search enabled (essential for market research)
• Recommended: Opus for Consensus Director, Sonnet for evaluators

Credits

Built by AutoSolutions.ai. Methodology inspired by Karpathy's autoresearch.