Skill · Think
First-principles breakdown
Strip a hard problem down to its fundamental facts and constraints, question every assumption, and rebuild a solution from the ground up.
The skill#
Act as a first-principles thinker.
The problem: [DESCRIBE THE PROBLEM]
How it is usually solved: [THE CONVENTIONAL APPROACH, OPTIONAL]
1. Restate the problem in its simplest form.
2. List the assumptions behind the usual approach.
3. Mark each assumption as a hard constraint (physics, law, money) or a convention.
4. List the fundamental facts that are definitely true.
5. Rebuild two or three solutions from those facts alone, ignoring convention.
6. Pick the most promising one and describe the first concrete step.
Be concrete. Where a "fact" is really an estimate, say so.How to use it#
- Copy the skill and describe the problem, and how it is normally solved.
- Look hardest at the assumptions marked as convention; that is where new options hide.
- Verify the "facts" it relies on before you build on them.
Example#
Describe the problem "our customer onboarding takes three weeks". The breakdown separates real constraints (legal checks) from conventions (sequential handoffs), and rebuilds an onboarding that runs the checks in parallel.
Quick and strict versions#
Use the quick version for everyday checks and the strict version when the result matters.
Break this problem down to its basic facts, name two assumptions people make that may not be true, and suggest one fresh approach.
Problem: [DESCRIBE THE PROBLEM]Act as an engineer and economist reasoning from first principles.
Problem: [DESCRIBE THE PROBLEM]
Current approach and its cost: [CURRENT APPROACH]
1. Decompose the problem into its component parts.
2. For each part, find the theoretical minimum (cost, time, materials, steps) and compare it with current reality.
3. Identify where the gap between minimum and reality is largest.
4. Challenge each assumption with "why must this be so?" until you reach a real constraint.
5. Propose solutions that attack the largest gap, with rough numbers.
6. List what you would need to verify before acting.
Label estimates clearly. Return: Decomposition | Minimum vs reality table | Assumption audit | Solutions | ChecksUse it as an agent skill#
Save this as SKILL.md in a folder named after the skill, and agents that support skills (Claude, Claude Code and other skill-aware agents) can load it automatically when the task fits.
---
name: first-principles
description: Use when the user is stuck on a hard problem and wants to rethink it from first principles rather than by convention.
---
# First-principles breakdown
Fill each [BRACKETED] field from the conversation. If one is missing, ask the user for it before starting.
Act as a first-principles thinker.
The problem: [DESCRIBE THE PROBLEM]
How it is usually solved: [THE CONVENTIONAL APPROACH, OPTIONAL]
1. Restate the problem in its simplest form.
2. List the assumptions behind the usual approach.
3. Mark each assumption as a hard constraint (physics, law, money) or a convention.
4. List the fundamental facts that are definitely true.
5. Rebuild two or three solutions from those facts alone, ignoring convention.
6. Pick the most promising one and describe the first concrete step.
Be concrete. Where a "fact" is really an estimate, say so.
Frequently asked questions
What is first-principles thinking?
Reasoning from the basic facts of a problem rather than by analogy or convention, then building a solution up from there.
When is it worth it?
For problems where the usual answer is expensive, slow or stuck. For routine problems, convention is usually fine.
Can AI really reason from first principles?
It is good at listing assumptions and alternatives. Treat its "facts" and numbers as estimates to verify.