Agent skill

first-principles

Stars 0
Forks 0

Install this agent skill to your Project

npx add-skill https://github.com/Jonely235/claude-skills/tree/main/.claude/skills/first-principles

SKILL.md

First-Principles Thinking Framework

When to Use This Skill

Apply this framework when you need to think independently rather than accepting:

  • Expert opinions or authority arguments
  • Conventional wisdom or "common knowledge"
  • Social proof or majority consensus
  • Emotional appeals or charismatic rhetoric
  • Historical precedent or tradition

The First-Principles Framework

Step 1: Identify the Question or Problem

State the core question explicitly:

  • What are we actually trying to understand?
  • What decision needs to be made?
  • What truth are we seeking?

Example: Instead of "Should I invest in crypto?", reframe as "What determines the value of a medium of exchange?"

Step 2: Extract All Assumptions

List every assumption in the argument:

  • What premises are being accepted without evidence?
  • What is being treated as "obvious" or "common knowledge"?
  • What authorities are being appealed to?
  • What emotional triggers are present?

Assumption Detection Patterns:

  • "Everyone knows..." → Flag as social proof
  • "Experts say..." → Flag as authority appeal
  • "It's obvious that..." → Flag as hidden premise
  • "Studies show..." → Demand methodology
  • "History proves..." → Demand context

Step 3: Strip to Fundamentals

Question each assumption:

  • Is this assumption actually true? How do we know?
  • Can we prove this from first principles?
  • What evidence would falsify this?
  • Is this a causal relationship or correlation?

Fundamental Truth Categories:

  • Logical truths: Mathematical/logical necessities (2+2=4)
  • Empirical truths: Observable, testable facts (water boils at 100°C at 1 atm)
  • Definitional truths: True by definition (bachelors are unmarried)
  • Reject: Tradition, authority, popularity, intuition

Step 4: Reconstruct from Ground Up

Build logical chains from proven fundamentals:

  1. Start with verified fundamental truths
  2. Apply valid deductive reasoning
  3. Test conclusions against empirical reality
  4. Identify what follows necessarily vs. probabilistically

Logical Validity Check:

  • Does conclusion follow necessarily from premises?
  • Are there logical fallacies (ad hominem, straw man, etc.)?
  • Are there missing premises or leaps in logic?
  • Is the argument structurally sound?

Step 5: Test Against Reality

Validate with empirical evidence:

  • What predictions does this reasoning generate?
  • Can we test those predictions?
  • What would count as falsification?
  • Are there counter-examples?

Bayesian Updating:

  • Assign initial probability based on first-principles reasoning
  • Update based on empirical evidence
  • Distinguish between "proven" and "more likely than not"

Step 6: Identify Remaining Uncertainty

Acknowledge what we don't know:

  • What premises remain uncertain?
  • What empirical evidence is missing?
  • What probability ranges are reasonable?
  • What would change our conclusion?

Intellectual Honesty:

  • Distinguish between "proven" and "probable"
  • Admit when evidence is insufficient
  • Identify what would settle the question
  • Avoid false certainty

Common Thinking Patterns to Avoid

Authority Worship

Pattern: "Expert X says Y, so Y is true" First-Principles Response: Authority is irrelevant to truth. Evaluate the argument, not the source.

Social Proof

Pattern: "Everyone believes X, so X is true" First-Principles Response: Majority opinion proves popularity, not truth. History is full of widely-believed falsehoods.

Tradition Appeal

Pattern: "We've always done X, so X is right" First-Principles Response: Past practice proves nothing about present optimality. Evaluate from first principles.

Confirmation Bias

Pattern: Seeking evidence that confirms existing beliefs First-Principles Response: Actively seek disconfirming evidence. Ask "What would prove me wrong?"

False Dichotomy

Pattern: "Either X or Y must be true" First-Principles Response: Are there other options? Are X and Y mutually exclusive? Question the frame.

Output Format

When applying this framework, structure responses as:

## [Core Question]

### Assumptions Identified
- [List all assumptions extracted]

### Fundamental Truths
- [List verified first principles]

### First-Principles Analysis
[Step-by-step logical reconstruction]

### Empirical Tests
[How to validate against reality]

### Remaining Uncertainty
[What we still don't know]

### Independent Conclusion
[Your conclusion based on above, not on others' opinions]

Key Principles

  1. Question Everything: No premise is accepted without justification
  2. Authority is Irrelevant: Truth exists independent of who says it
  3. Logic Over Intuition: Feelings are not evidence
  4. Evidence Over Assertion: Demand proof, don't accept claims
  5. Uncertainty is Honesty: Admit what you don't know
  6. Update Beliefs: Change your mind when evidence demands it

Advanced Frameworks

For complex problems, see:

  • frameworks.md: Detailed methods (5 Whys, Socratic questioning, assumption mapping)
  • examples.md: Concrete applications across domains

Remember: The goal is not to be contrarian, but to be correct. First-principles thinking often leads to unconventional conclusions because conventional thinking is often flawed.

Expand your agent's capabilities with these related and highly-rated skills.

Jonely235/claude-skills

surgical-injector

Safely inject new code into legacy codebases with risk containment. Use for feature flags, gradual rollouts, A/B testing new logic, or refactoring critical paths. Implements try-catch fallbacks, feature toggle architecture, and surgical separation of old/new code.

0 0
Explore
Jonely235/claude-skills

surgical-injector

Safely inject new code into legacy codebases with risk containment. Use for feature flags, gradual rollouts, A/B testing new logic, or refactoring critical paths. Implements try-catch fallbacks, feature toggle architecture, and surgical separation of old/new code.

0 0
Explore
Jonely235/claude-skills

blast-radius-analysis

Analyze the impact of code changes by mapping call graphs and identifying all direct and indirect dependencies. Use when users ask about blast radius analysis, code change impact, who calls a function, downstream dependencies, or risk assessment before refactoring. Traces call chains, identifies high-risk modules, and provides ranked impact analysis.

0 0
Explore
Jonely235/claude-skills

code-cartographer

Analyze and map code structure, data flow, and dependencies in legacy codebases. Use for understanding call stacks, dependency trees, side effects, risk profiles, and generating Mermaid diagrams. Ideal for legacy codebases with zero tests.

0 0
Explore
Jonely235/claude-skills

skill-creator

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

0 0
Explore
Jonely235/claude-skills

code-cartographer

Analyze and map code structure, data flow, and dependencies in legacy codebases. Use for understanding call stacks, dependency trees, side effects, risk profiles, and generating Mermaid diagrams. Ideal for legacy codebases with zero tests.

0 0
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results