# Lean Startup Validation

> Run a Lean Startup validation loop with assumptions, MVP, build-measure-learn, validated learning, and pivot/persevere decisions. Use when Codex is asked to test a startup, product, feature, market, or business model under uncertainty.

- Skill: `walterkken/lean-startup-validation` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add walterkken/lean-startup-validation`
- Raw SKILL.md: https://api.skillmd.com/api/skills/walterkken/lean-startup-validation/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: walterkken (https://skillmd.com/u/walterkken)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/walterkken/lean-startup-validation

---


# Lean Startup Validation

## Workflow

1. State the riskiest assumption.
2. Define what learning would change the decision.
3. Build the smallest MVP or experiment that tests the assumption.
4. Measure behavior, not opinions, when possible.
5. Learn: compare data against the decision threshold.
6. Decide: persevere, pivot, stop, or run another experiment.

## 马斯克 Supervision

Delete features that do not test the riskiest assumption. Do not automate or scale before validated learning.

## Output

- assumption map
- MVP/experiment design
- metric and threshold
- learning report
- pivot/persevere decision
- next experiment

See `references/source-notes.md`.