first-principles

name	first-principles
description	Break down assumptions and rebuild reasoning from fundamental truths — use when a standard approach feels wrong or produces diminishing returns but no one has questioned its underlying assumptions
argument-hint	[problem or system to deconstruct]
user-invocable	true
license	MIT

Deconstruct a problem into its most basic, foundational elements. Strip away assumptions, conventions, and analogies. Rebuild understanding from verified fundamentals to find novel solutions. 1. State the problem or system clearly in one sentence 2. List every assumption embedded in the current approach 3. Challenge each assumption — is it a fundamental truth or a convention? 4. Identify the bedrock truths that remain after challenges 5. Rebuild a solution from only those verified fundamentals 6. Compare the rebuilt solution to the conventional approach 7. Identify what the conventional approach gets wrong or misses

<output_format>

First-Principles Analysis: [Topic]

Problem Statement

[One clear sentence]

Assumptions Identified

• [Assumption 1] — fundamental / convention / unverified
• [Assumption 2] — ...

Fundamental Truths

1. [Verified truth that survived challenge]
2. ...

Rebuilt Reasoning

[Solution constructed from fundamentals only]

Key Insight

[What changes when you reason from fundamentals vs convention] </output_format>

## First-Principles Analysis: Monolith vs Microservices

Problem Statement

Our team assumes we need microservices because we're scaling, but deploys are slow and coordination costs are high.

Assumptions Identified

• Microservices improve scalability — convention (monoliths scale vertically too)
• Independent deploys are faster — convention (coordination overhead can negate this)
• Our team is big enough to own separate services — unverified (we have 4 engineers)
• Network calls between services are negligible — unverified (latency adds up)

Fundamental Truths

1. We need to deploy changes without breaking unrelated features
2. Our current bottleneck is deploy pipeline speed, not runtime scaling
3. Four engineers cannot maintain more than 2-3 independent services effectively

Rebuilt Reasoning

A modular monolith with clear internal boundaries gives us deploy isolation (via feature flags and targeted rollouts) without the coordination cost of service boundaries. Split only when a module's scaling needs diverge measurably from the rest.

Key Insight

The assumption "scaling = microservices" skipped the question "what are we actually scaling?" Our bottleneck is deploy speed, not request throughput — a problem microservices make worse, not better.

Key Instructions

• If all assumptions survive challenge, note that the conventional approach may already be optimal — first-principles doesn't always produce a novel solution.

Anti-Pattern Guards

1. Rebuilding from convention — stripping away a convention only to rebuild the same approach is the common failure; the rebuilt solution must trace to verified fundamentals, not intuition.
2. Mistaking "feels fundamental" for "is fundamental" — a fundamental truth is verifiable, not just hard to argue with; apply the same skepticism to proposed fundamentals as to assumptions.

Handoff

Chainable to: inversion (to stress-test the rebuilt solution), consider (to apply additional mental models)