Validate Suite — Development skill for Claude Code

validate-suite

**A decision library for your AI agents.** Point it at any idea, feature, pricing, or GTM decision and it runs the case through a **9-stage, evidence-gated pipeline** — SWOT, Five Forces, premortem, red-team and nine other critique lenses — then hands you a plain-language HTML dossier ending in **Go / No-go / Pivot / Park**, with an honest confidence score and concrete coaching.

**🌐 Language:** [English](#english) · [Tiếng Việt](#tiếng-việt)

---

English

Who it's for

• Solo founders with no board to bounce ideas off — get a second opinion that always asks "where's the evidence?"
• Product managers defending a roadmap — replace "I think we should build X" with a defensible, scored case
• Anyone pre-GTM — stress-test pricing, a new channel, or a beachhead before spending money on it
• Small-business owners & consultants — one repeatable process instead of ad-hoc analysis for every big call

Why not just ask your AI "should we do this?"

Because you'll get a fluent, confident essay — and AI models are known to invent plausible numbers (TAM, CAC, conversion) when none exist. `validate-suite` exists to prevent exactly that:

• Evidence discipline — every score is tagged with where it came from, and weak evidence automatically caps scores and drags confidence down. A confident "GO" can never rest on guesses.
• Gates, not vibes — cheap binary gates kill bad ideas in minutes, before any expensive analysis.
• Structured critique instead of one voice — Six Thinking Hats, a multi-persona debate, premortem, inversion, red-team, weighted matrix, SWOT, Porter's Five Forces.
• Self-verifying report — the HTML recalculates every score and confidence in your browser and flags any mismatch, so the numbers you read are numbers that add up.

How a run works

1. Say it naturally: "should we build feature X this sprint?" · "validate ý tưởng X" · "stress-test this pricing change" · *"sho