スキル一覧に戻る

first-principles

guia-matthieu
更新日 2 days ago
4 閲覧
111
20
111
GitHubで表示
その他general

について

このスキルは、第一原理思考を実装し、複雑な問題を核心的真実へと解体し、革新的な解決策をゼロから再構築します。従来の手法が通用せず、基本的な公理から推論する必要がある場合に、開発者が一見不可能な課題に取り組み、業界の前提を打ち破り、抜本的な製品革新を必要とする場面で活用するために設計されています。

クイックインストール

Claude Code

推奨
メイン
npx skills add guia-matthieu/clawfu-skills -a claude-code
プラグインコマンド代替
/plugin add https://github.com/guia-matthieu/clawfu-skills
Git クローン代替
git clone https://github.com/guia-matthieu/clawfu-skills.git ~/.claude/skills/first-principles

このコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします

ドキュメント

First Principles Thinking

Break down complex problems to their fundamental truths, then reason up from there. Master Aristotle's ancient method modernized by Elon Musk to solve seemingly impossible problems.

When to Use This Skill

  • Facing a "impossible" problem where conventional solutions don't work
  • Challenging industry assumptions that everyone accepts as truth
  • Pricing or cost analysis to find dramatic savings (like Musk with batteries)
  • Product innovation when iterating on existing solutions isn't enough
  • Strategic decisions where analogies to other companies may mislead
  • Breaking through mental blocks when you're stuck in conventional thinking

Methodology Foundation

AspectDetails
SourceAristotle (384-322 BC), modernized by Elon Musk
ExpertAristotle called it "the first basis from which a thing is known"; Musk applies it to SpaceX, Tesla, and every venture
Core Principle"Boil things down to the most fundamental truths and say, 'What are we sure is true?'... and then reason up from there." - Elon Musk

What Claude Does vs What You Decide

"Claude structures the breakdown. You challenge the assumptions."

Claude handlesYou provide
Systematically listing assumptionsDomain expertise to validate/invalidate
Breaking down atomic componentsReal-world cost/constraint data
Applying the physics vs convention testJudgment on what's truly immutable
Structuring the rebuild from fundamentalsStrategic decisions on which path to pursue
Generating novel solution frameworksFeasibility assessment and implementation

Remember: First principles is a thinking tool. Claude accelerates the structure; breakthrough insights come from YOUR willingness to question everything.


What This Skill Does

  1. Deconstructs assumptions - Identifies what you think is true vs. what IS true
  2. Reveals hidden constraints - Exposes artificial limitations masquerading as laws
  3. Enables breakthrough solutions - Creates paths invisible to analogical thinkers
  4. Reduces costs dramatically - Finds the true floor price of anything
  5. Builds original insights - Generates conclusions others can't reach by copying

How to Use

Challenge Conventional Wisdom

Apply First Principles Thinking to challenge this assumption:
[describe the "truth" everyone accepts in your industry]

Break it down to fundamental truths and show me what's actually possible.

Analyze Costs or Pricing

Use First Principles to analyze the true cost of:
[product, service, or process]

What are the raw components? What should this actually cost?

Solve an "Impossible" Problem

I'm told [problem] is impossible because [reasons].
Apply First Principles to find a path forward.
What fundamental truths apply here? What assumptions can we challenge?

Instructions

When applying First Principles Thinking, follow this systematic process:

Step 1: Identify the Problem and Current Assumptions

## Current State Analysis

**Problem:** [What are you trying to solve?]

**Current "Truth":** [What does everyone believe about this?]

**Analogies in Use:** [How do people currently reason about this?]
- "It's always been done this way"
- "Company X does it like this"
- "Industry standard is..."

**Hidden Assumptions:** (List everything assumed to be true)
1.
2.
3.
4.
5.

Key Questions:

  • What do "experts" say is impossible or fixed?
  • What have you never questioned about this problem?
  • What would a complete outsider find strange about your assumptions?

Step 2: Break Down to Fundamental Truths

## First Principles Breakdown

For each assumption, ask: "Is this a LAW OF NATURE or a CONVENTION?"

| Assumption | Type | Evidence |
|------------|------|----------|
| [Assumption 1] | Law / Convention | [Why?] |
| [Assumption 2] | Law / Convention | [Why?] |
| [Assumption 3] | Law / Convention | [Why?] |

**Laws of Nature** (Cannot be changed):
- Physics (gravity, thermodynamics, etc.)
- Mathematics (proven theorems)
- Biology (human needs)

**Conventions** (CAN be changed):
- Industry practices
- "Best practices"
- Historical precedent
- Regulatory (can be challenged)
- Social norms

The Physics Test: Ask: "Does the law of physics prevent this?" If no, it's changeable.

Elon Musk's 3-Step Process:

  1. Identify and define current assumptions
  2. Break down the problem into fundamental principles
  3. Create new solutions from scratch

Step 3: Find the Atomic Components

## Atomic Breakdown

**What is this thing MADE OF at the most basic level?**

Example: Battery Pack
├── Lithium: $X per kg
├── Cobalt: $X per kg
├── Nickel: $X per kg
├── Aluminum: $X per kg
├── Steel: $X per kg
├── Polymers: $X per kg
└── Manufacturing energy: $X

**Raw Material Cost:** $___
**Current Market Price:** $___
**Gap to Explain:** $___

The gap represents: convention, inefficiency, margin, or artificial scarcity

For Non-Physical Problems:

  • What are the atomic ACTIONS required?
  • What are the atomic INPUTS needed?
  • What are the atomic CONSTRAINTS (true constraints)?

Step 4: Reason Up from Fundamentals

## Rebuild from First Principles

Starting ONLY from verified truths, ask:

1. **What is physically possible?**
   [Based on laws of nature]

2. **What is the theoretical minimum?**
   [Cost, time, resources, steps]

3. **What new combination could work?**
   [Ignoring how it's "supposed" to be done]

4. **What would we do if starting from scratch?**
   [No legacy constraints]

## New Solution Design

**Fundamental Requirements:**
- [Only TRUE requirements]

**Eliminated "Requirements":**
- [Things assumed necessary but aren't]

**Novel Approach:**
[Description of solution built from first principles]

Step 5: Validate and Iterate

## Validation Checklist

- [ ] Solution respects actual laws of nature
- [ ] No hidden conventional assumptions snuck back in
- [ ] Can explain WHY each element is necessary
- [ ] Theoretical minimum is understood
- [ ] Gap between current and theoretical is explained

## Implementation Path

**What makes this hard (real obstacles):**
1.

**What makes this SEEM hard (false obstacles):**
1.

**First experiment to validate:**
[Smallest test of the core assumption]

Examples

Example 1: SpaceX Rocket Cost Reduction

Context: Elon Musk wanted to buy rockets for Mars mission. Quote: $65 million per rocket.

Input:

"Everyone says rockets cost $65M and that's just what they cost. Is this true? Apply First Principles."

Output:

Current State Analysis:

AssumptionChallenge
Rockets are expensiveConvention - based on historical pricing
Must buy from established vendorsConvention - could manufacture
Rockets are single-useConvention - airplanes aren't single-use
Aerospace requires premium pricingConvention - not physics

First Principles Breakdown:

"What is a rocket made of?"

  • Aerospace-grade aluminum
  • Titanium
  • Copper
  • Carbon fiber
  • Fuel (kerosene, liquid oxygen)

Atomic Cost Analysis:

Raw materials for Falcon 9:
├── Aluminum alloys: ~$35,000
├── Titanium: ~$15,000
├── Carbon fiber: ~$20,000
├── Avionics components: ~$100,000
├── Engines (9x Merlin): ~$1M (at scale)
└── Other materials: ~$30,000

Theoretical material cost: ~$1.2M
Industry price: $65M
Gap: 54x markup

Reasoning Up:

  1. Physics allows: Rockets can be reusable (planes land and fly again)
  2. Minimum cost: ~$1.2M in materials + manufacturing + margin = $5-10M possible
  3. Novel approach:
    • Manufacture in-house (eliminate contractor margins)
    • Design for reusability (eliminate single-use assumption)
    • Vertical integration (control supply chain)

Result: Falcon 9 now costs ~$28M to launch, with reusability bringing marginal cost to ~$15M. Other companies still charge $100M+.

Why This Works: Musk didn't accept "rockets cost $65M" as a truth. He found it was a convention built on inefficient practices, not physics.


Example 2: Rethinking Employee Training Costs

Context: Company spends $2,000 per employee on annual training. CFO says "that's industry standard."

Input:

"We spend $2,000/employee on training. HR says that's normal. Is it? First Principles analysis please."

Output:

Current State Analysis:

AssumptionType
Training must be instructor-ledConvention
Must use external vendorsConvention
Must be synchronousConvention
Certifications require paid coursesConvention
Training = classroom timeConvention

Atomic Breakdown - What does training actually require?

Training = Transfer of Knowledge + Skill Practice + Verification

├── Knowledge Transfer
│   ├── Content (can be recorded once)
│   ├── Delivery (can be async/digital)
│   └── Q&A (can be AI-assisted or peer)
│
├── Skill Practice
│   ├── Exercises (can be self-paced)
│   ├── Feedback (can be automated or peer)
│   └── Repetition (individual, no instructor needed)
│
└── Verification
    ├── Testing (automated)
    └── Observation (sampling, not 100%)

Cost Breakdown:

Current $2,000/person:
├── Vendor fees: $800 (40%)
├── Instructor time: $500 (25%)
├── Travel/facilities: $400 (20%)
├── Materials: $200 (10%)
└── Admin overhead: $100 (5%)

First Principles minimum:
├── One-time content creation: $50/person (amortized)
├── Platform/LMS: $30/person/year
├── Peer mentoring time: $100/person
├── Verification tools: $20/person
└── Buffer: $50/person

Theoretical minimum: ~$250/person

Reasoning Up - New Model:

  1. Create content once: Record best trainer explaining each topic. Amortize over 5 years.
  2. Async delivery: Employees learn at their pace, no scheduling costs
  3. AI Q&A: Claude/GPT handles 80% of questions
  4. Peer practice: Employees practice together (free)
  5. Automated verification: Digital assessments, spot-check practical

New Solution:

  • Year 1: $500/person (content creation investment)
  • Year 2+: $250/person (75% reduction maintained)

Validation:

  • Still transfers knowledge? Yes
  • Still builds skills? Yes (peer practice)
  • Still verifies? Yes (automated + sampling)
  • Respects laws of learning? Yes

Checklists & Templates

First Principles Analysis Template

## First Principles Analysis: [Topic]

### 1. THE ASSUMPTION AUDIT

**What "everyone knows" about this:**
1.
2.
3.

**Source of these beliefs:**
- [ ] Physics/Math (unchangeable)
- [ ] Biology (unchangeable)
- [ ] Regulation (changeable but hard)
- [ ] Industry practice (changeable)
- [ ] Historical precedent (changeable)
- [ ] Someone told me (verify)

### 2. THE ATOMIC BREAKDOWN

**Physical components:**
- Component 1: [cost/quantity]
- Component 2: [cost/quantity]
- ...

**Required actions (minimum):**
1.
2.
3.

**True constraints (laws of nature):**
-

**False constraints (conventions):**
-

### 3. THE REBUILD

**If we started from scratch with only true constraints:**

**Theoretical minimum:** [cost/time/resources]

**Current state:** [cost/time/resources]

**Gap:** [X%]

**Gap explained by:**
- [ ] Inefficiency
- [ ] Margin/profit
- [ ] Convention
- [ ] Lack of innovation
- [ ] True complexity (irreducible)

### 4. THE NEW SOLUTION

**Core insight:**

**New approach:**

**First experiment to test:**

**What could go wrong:**

Quick First Principles Checklist

## Before Accepting Any "Truth"

□ Is this a law of physics/math/biology?
□ Or is it "how things are done"?
□ Who benefits from this being "true"?
□ What would a complete outsider think?
□ Has anyone ever done it differently?
□ What would this look like if we started fresh?

## The Musk Test

□ Can I explain the atomic components?
□ Do I know the theoretical minimum cost?
□ Is there a 10x gap I can't explain?
□ Am I reasoning by analogy or from fundamentals?

Common First Principles Domains

## Where to Apply First Principles

### COST REDUCTION
- What are the raw materials?
- What is labor actually required?
- What steps can be eliminated?

### PRODUCT DESIGN
- What does the user fundamentally need?
- What is convention vs. requirement?
- What would this look like in 10 years?

### PROCESS IMPROVEMENT
- What are the required steps (physics)?
- What are the assumed steps (convention)?
- What is the theoretical minimum time?

### PRICING
- What is the cost to deliver?
- What is convention in pricing?
- What would a new entrant charge?

### STRATEGY
- What does the customer fundamentally want?
- What is assumed about competition?
- What constraints are real vs. imagined?

Red Flags: When You're NOT Using First Principles

## Warning Signs

You're reasoning by ANALOGY (not First Principles) when you say:

- [ ] "Industry standard is..."
- [ ] "Competitors do it this way..."
- [ ] "We've always done it like this..."
- [ ] "Best practice says..."
- [ ] "Experts recommend..."
- [ ] "It's common knowledge that..."
- [ ] "That's just how it works..."

None of these are REASONS. They're appeals to convention.

## The Fix

For each statement above, ask:
"But WHY? What fundamental truth makes this necessary?"

If you can't answer with physics, math, or biology - it's changeable.

Skill Boundaries (Frontier Recognition)

This skill excels for:

  • Cost reduction analysis (finding the "true floor")
  • Challenging industry assumptions everyone accepts
  • Product innovation beyond incremental improvements
  • Strategic decisions where analogies may mislead
  • Pricing strategy (understanding what things SHOULD cost)

This skill is NOT ideal for:

  • Quick decisions where speed matters more than depth → Use heuristics instead
  • Execution tasks where the path is already clear → First principles is overkill
  • Emotional/relationship problems → Not everything reduces to physics
  • Regulatory constraints that truly can't be changed → Accept and work within
  • Pure creative work where logic isn't the bottleneck → Use different creative tools

Quality Checkpoints

Before accepting the analysis, verify:

  • You've genuinely identified the REAL alternatives (not just direct competitors)
  • "Laws of physics" are actually physics, not just hard conventions
  • The atomic cost breakdown uses real market data
  • Novel solution doesn't sneak conventional assumptions back in
  • You have a way to validate the first insight before committing

Iteration Guide

"First principles is a dialogue, not a one-shot analysis."

Recommended Iteration Pattern

PassFocusQuestions to Ask
1stAssumption dump"List every assumption in this problem"
2ndChallenge"For each, is it physics or convention?"
3rdAtomic"What is this actually made of / what does it actually require?"
4thRebuild"Starting from only true constraints, what's possible?"

Useful Follow-up Prompts

After the first analysis, try:

  • "I think [assumption] is actually physics because [reason]. Convince me otherwise."
  • "The atomic breakdown is missing [component]. Redo with this included."
  • "Everyone accepts [constraint] as given. Push harder on whether it's truly immutable."
  • "The 'novel solution' still feels conventional. What would be a 10x different approach?"

Learning Curve

UsageWhat You'll Experience
1st useStructured assumption breakdown, discover the method
3rd useYou catch yourself reasoning by analogy and stop
10th useFirst principles becomes a reflex when facing "impossible" problems

Pro tip: The biggest gains come when you're willing to look stupid by questioning "obvious" truths. If everyone already knows it's changeable, it's not a first principles insight.


References

  • Aristotle. "Posterior Analytics" (~350 BC) - Original first principles concept
  • Musk, Elon. Various interviews on SpaceX methodology (2012-present)
  • Clear, James. "First Principles: Elon Musk on the Power of Thinking for Yourself"
  • Farnam Street. "First Principles: The Building Blocks of True Knowledge"
  • Munger, Charlie. "Elementary Worldly Wisdom" - Adjacent mental model thinking

Related Skills


Skill Metadata

  • Mode: cyborg
name: first-principles
category: strategy
subcategory: problem-solving
version: 2.0
author: GUIA
source_expert: Aristotle, Elon Musk
source_work: Posterior Analytics (Aristotle), SpaceX methodology (Musk)
difficulty: intermediate
mode: both  # Both = can be structured (Centaur) or exploratory (Cyborg) depending on use
estimated_value: $2,000 strategy consulting session
tags: [problem-solving, innovation, cost-reduction, assumptions, Musk, thinking]
created: 2026-01-25
updated: 2026-01-28

This skill is part of the GUIA Premium Marketing Skills Library — the 201 layer that bridges AI basics and technical implementation.

GitHub リポジトリ

guia-matthieu/clawfu-skills
パス: skills/strategy/first-principles
0
ai-skillsanthropicclaude-codeclaude-skillsmarketingmcp-server

関連スキル

llamaguard

その他

LlamaGuardは、暴力やヘイトスピーチなど6つの安全性カテゴリーにおいて、LLMの入力と出力をモデレートするMetaの70-80億パラメータモデルです。94〜95%の精度を提供し、vLLM、Hugging Face、Amazon SageMakerを使用してデプロイ可能です。このスキルを使用して、AIアプリケーションにコンテンツフィルタリングと安全策を簡単に統合できます。

スキルを見る

cost-optimization

その他

このClaudeスキルは、リソースの適正サイジング、タグ付け戦略、支出分析を通じて、開発者がクラウドコストを最適化することを支援します。AWS、Azure、GCPにわたるクラウド支出の削減とコストガバナンスの実施のためのフレームワークを提供します。インフラコストの分析、リソースの適正サイジング、または予算制約への対応が必要な際にご利用ください。

スキルを見る

quantizing-models-bitsandbytes

その他

このスキルは、bitsandbytesを使用してLLMを8ビットまたは4ビット精度に量子化し、精度の低下を最小限に抑えつつ50〜75%のメモリ削減を実現します。限られたGPUメモリでより大規模なモデルを実行したり、推論を高速化するのに理想的で、INT8、NF4、FP4などのフォーマットをサポートしています。HuggingFace Transformersと統合され、QLoRAトレーニングや8ビットオプティマイザーを可能にします。

スキルを見る

dispatching-parallel-agents

その他

このClaudeスキルは、複数のエージェントを配備し、3つ以上の独立した問題を並行して調査・修正します。共有状態や依存関係がなく解決可能な、無関係な障害が発生するシナリオ向けに設計されています。中核となる機能は並列問題解決であり、効率を最大化するために独立した問題領域ごとに1つのエージェントを割り当てます。

スキルを見る