Skill: Validate Concept

Apply the Meta-Validation Framework to any new product concept before investing strategy/engineering effort. Produces a 1-page viability scorecard with GO/CAUTION/KILL verdict.

Trigger

User says: "validate concept", "validar idea", "run the framework on X", "is this viable", "should we build this", "kill or keep".

Purpose

Prevent the Tesla Fleet failure mode: writing a 269-line strategy doc before anyone checked whether the core technical premise was true. The framework exists to catch that on turn 1, not turn 20.

Source

Framework defined in .unused/future/META_VALIDATION_FRAMEWORK.md (or vault/19-validation/framework/meta-validation-framework.md after migration). Read it first.

Workflow

Step 1: Identify the Concept

Get from the user:

• Concept name
• One-line pitch (≤140 chars)
• What it claims to do
• What infrastructure it depends on

Step 2: Check Against 5 Failure Modes (F1-F5)

For each, ask yes/no:

Any YES = high risk. Proceed with extra scrutiny.

Step 3: Define the 3 Kill-Switches

Every concept must have 3 concrete yes/no questions whose NO answer kills it immediately.

Template:

Kill-Switch #1 (Technical): Does [primitive X] actually expose [capability Y] today?
  - NO path: Concept is dead unless [specific alternative exists]
  - Evidence required: Direct quote from API docs or test call
  - Owner: Tech-Reality Auditor

Kill-Switch #2 (Economic): Will [buyer type Z] pay at least $[amount] per [unit]?
  - NO path: Concept is unprofitable
  - Evidence required: 1 signed LOI, 3 buyer quotes, or equivalent existing contract
  - Owner: Unit Economist

Kill-Switch #3 (Legal or Gatekeeper): Can we run this without [blocker's permission]?
  - NO path: Concept depends on a gatekeeper who can say no
  - Evidence required: Written legal opinion or ToS reading
  - Owner: Legal Scanner

Write these BEFORE doing any research. If you can't write them concretely, the concept isn't ready.

Step 4: Score Against 6 Axes

Step 5: Produce the 1-Page Scorecard

Generate the scorecard in this exact format and save to vault/19-validation/scorecards/[concept-slug]-scorecard.md:

---
date: YYYY-MM-DD
type: concept-scorecard
status: draft | validated | killed
concept: [name]
verdict: GO | CAUTION | KILL
related:
  - "[[concept-file]]"
  - "[[meta-validation-framework]]"
tags:
  - type/scorecard
  - domain/validation
---

# [Concept Name] — Viability Scorecard

**One-line pitch:** [≤140 chars]

## The 3 Kill-Switches
1. [technical question] — [ Y / N / UNKNOWN ]
2. [economic question] — [ Y / N / UNKNOWN ]
3. [legal/gatekeeper question] — [ Y / N / UNKNOWN ]

## 6-Axis Scorecard
- **A1. Technical Reality:** [ PASS / CAUTION / KILL ] — [reason]
- **A2. Legal & Regulatory:** [ PASS / CAUTION / KILL ] — [reason]
- **A3. Data Quality:** [ PASS / CAUTION / KILL ] — [reason]
- **A4. Economics:** [ PASS / CAUTION / KILL ] — [reason]
- **A5. Partnership/GTM:** [ PASS / CAUTION / KILL ] — [reason]
- **A6. Buyer Demand:** [ PASS / CAUTION / KILL ] — [reason]

## Top 3 Risks
1. [risk] — [likelihood×impact score] — [mitigation or "kill switch"]
2. [risk] — [score] — [mitigation]
3. [risk] — [score] — [mitigation]

## First Real Customer
[person/company] — [contact path] — [why them]

## What EM Already Has (% reuse)
[e.g. 70% — escrow, evidence pipeline, ERC-8004, worker pool]

## Verdict
**[GO / CAUTION / KILL]**

**Next step:** [concrete action with owner and deadline]

Step 6: Apply Verdict Rules

• GO = 6/6 PASS, or 5 PASS + 1 CAUTION with mitigation plan → proceed to full deep-dive
• CAUTION = 4-5 PASS, no KILL → focused research sprint on CAUTION axes before committing
• KILL = Any 1 KILL on any axis → archive to vault/21-graveyard/ with post-mortem

Step 7: Notify User

Print in 3 lines max:

Scorecard saved: [[concept-name-scorecard]]
Verdict: [GO / CAUTION / KILL]
Next: [action]

Anti-Patterns (Do NOT Do)

• ❌ Fill the scorecard with optimism. Every PASS needs evidence.
• ❌ Skip the kill-switches. They're the whole point of the framework.
• ❌ Validate multiple concepts in one pass. One at a time.
• ❌ Accept "cities" or "insurance companies" as a first customer. Named or nothing.
• ❌ Write the strategy doc before the scorecard. Scorecard first, always.

Universal Insights (Must Apply)

1. "Infrastructure already exists" is usually a lie about the 10% that matters.
2. "Zero user effort" claims need adversarial review.
3. Write the scorecard BEFORE the strategy document.
4. Specialist disagreement is a feature.
5. "X already does Y, but we do Y+Z" is the highest-risk pitch class.
6. Name the first customer or kill the concept.