n-whys

Generate a why-chain exactly n levels deep for any topic, question, observation, problem, or idea. Invoke like `n-whys n=8 q=<my question>`. The general form of five-whys — use this when you want an e

SKILL.md

GitHub

N Whys

Use this skill when the user invokes n-whys or asks for a why-chain of a specified depth.

Relationship to five-whys: five-whys is a preset of this skill — depth ~5, root-cause orientation, flexible stopping. Use n-whys when the user wants a deliberate, exact depth or an unusually deep chain.

Invocation

Expected form:

n-whys n=<integer> q=<question or topic>

Examples:

n-whys n=8 q=Why do people procrastinate on work they care about?
n-whys n=4 q=Why is this product hard to explain?
n-whys n=10 q=AI agents feel brittle

Arguments

  • n: required integer depth for the chain.
  • q: required question, topic, statement, problem, or observation to analyze.

If n or q is missing, ask for the missing argument. If n is very large, proceed unless the user asks for something impractically long; keep each layer concise.

Purpose

N Whys generalizes the Five Whys. It repeatedly asks "why?" exactly n times to expose deeper causes, assumptions, mechanisms, incentives, constraints, or principles beneath a surface question.

The chain can take several forms depending on the topic:

  • causal — event chains, mechanisms, failures
  • conceptual — deeper premises behind a claim
  • incentive-based — who benefits, what pressures exist
  • psychological — motivations, fears, needs
  • philosophical — grounding values and assumptions

Canonical questions

  • Why do people procrastinate on work they care about? (n=8)
  • Why is the sky blue? (n=6, physics chain)
  • Why do startups die? (n=10)
  • Why does this abstraction feel wrong? (n=5)

Procedure

  1. Parse n and q from the invocation.
  2. Restate q as the surface question or topic.
  3. Produce exactly n numbered why-layers.
  4. Each layer should:
    • ask a natural "why" question about the previous layer,
    • answer it directly,
    • go one level deeper than the prior answer.
  5. Keep facts, hypotheses, and interpretations distinct.
  6. After the chain, synthesize the deepest insight or principle.
  7. End with implications: what this suggests doing, checking, reframing, or asking next.

Output format

## Topic
<q>

## Why chain
1. **Why ...?**
   - ...
2. **Why ...?**
   - ...
...
N. **Why ...?**
   - ...

## Deepest principle
...

## Implications
- ...

Guardrails

  • Produce exactly n why-layers unless the user changes the request.
  • Do not pad with fake certainty. If a link is speculative, label it as a hypothesis.
  • Avoid blame in human or organizational topics; prefer systems, incentives, constraints, and feedback loops.
  • If the chain becomes circular, name the loop and continue by asking why the loop persists.
  • If the topic is conceptual rather than causal, interpret "why" as "what deeper premise explains this?"