Skip to main content
skill CX Codex

thinking-first-principles — Claude Code skill: rebuild from facts

A skill that makes the agent separate real constraints from convention and rebuild the simplest solution. Useful when a task looks too costly or impossible.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add tjboudreaux/cc-thinking-skills --skill thinking-first-principles -a claude-code
Open on GitHub

What thinking-first-principles does

The skill kicks in when a limit is taken as given: "too expensive", "impossible", "we've always done it this way". The agent breaks the problem down to primitives and rebuilds the simplest thing that fits the real constraints.

It runs as a fixed procedure. First the agent lists every constraint, including the unspoken ones. Each gets a tag: physics or math, regulation or contract, measured fact, or convention and authority. The first three stay binding; the rest are dropped unless someone can back them with evidence. Then comes the rebuild from what's left, a check against real needs, and the cheapest test that would falsify the result.

Inside there is a single SKILL.md with the procedure and a section on when not to use it. A constraint backed by physics, law or measured capacity gets accepted, and you optimize inside it. A proven library or protocol also rules out a rebuild. During an incident you fix the likeliest cause first and save the redesign for afterwards.

What it helps with

  • tell real constraints from inherited assumptions
  • rebuild a solution once analogy stops working
  • check an estimate that rests on vendor pricing instead of a derivation
  • cut moving parts out of an architecture at the start
  • find a cheap test that would falsify the chosen approach

Example prompts for the agent

  • break this constraint down to first principles: what is physics here and what is convention
  • we're told this can't be done — check what the ban actually rests on
  • rebuild the solution from verified requirements only

How to install thinking-first-principles

1Skills CLI (Vercel)

npx skills add tjboudreaux/cc-thinking-skills --skill thinking-first-principles -a claude-code

2Manually: put the files in place

~/.claude/skills/thinking-first-principles/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/thinking-first-principles/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about thinking-first-principles

How do I install thinking-first-principles in Claude?

Run npx skills add tjboudreaux/cc-thinking-skills --skill thinking-first-principles -a claude-code. The skill file lands at ~/.claude/skills/thinking-first-principles/SKILL.md.

What does thinking-first-principles do?

It helps the agent decompose a problem into primitives, drop unsupported assumptions and rebuild a simple solution from what remains. The output covers tagged constraints, discarded conventions, the rebuild and a falsifying test.

When should I not use this skill?

When the constraint is verified physics, regulation or a measured limit, accept it and optimize within. A proven library or protocol also settles the question, and during an incident you fix the likely cause first.

Source text

SKILL.md · MIT
---
name: thinking-first-principles
description: When a constraint is treated as fixed, separate physics from convention, keep only independently supported primitives, and rebuild the simplest solution that satisfies real constraints.
disable-model-invocation: true
---

# First Principles

Strip assumed constraints to independently supported primitives, then rebuild. Escape analogy when convention fails or looks artificially expensive.

## When to Use

- A constraint is treated as fixed ("too expensive," "impossible," "always done this way") without independent support.
- Convention or analogy has failed, stalled, or only yields weak incremental variants.
- Greenfield design where industry defaults may smuggle false requirements.
- Cost or feasibility claims rest on pricing, precedent, or authority rather than physics, math, or measured need.

## When NOT to Use

- The constraint is verified physics, hard regulation, or measured capacity → accept it and optimize within it.
- A proven standard library, protocol, or known-good pattern already fits → use it; do not re-derive.
- Time-critical incidents → act on the most likely cause first; reserve first-principles rebuild for post-incident redesign.
- Incremental polish of a working system with no false-constraint signal → skip.

## Procedure

1. **State the problem and claimed constraints.** List every limit treated as fixed (cost, scale, stack, process, "must use X").
2. **Classify each constraint.** Tag as `physics/math`, `regulation/contract`, `measured fact`, or `convention/analogy/authority`. Keep only the first three as binding unless evidence upgrades a convention.
3. **Decompose to primitives.** Required inputs, conservation/complexity bounds, true non-negotiables. Demand independent support (measurement, derivation, primary requirement)—not vendor pricing or competitor precedent.
4. **Discard unsupported assumptions.** For each convention tag, state a falsifier. Drop or renegotiate anything lacking support.
5. **Rebuild from remaining primitives only.** Simplest solution that satisfies binding constraints and ignores artificial ones. Prefer commodity inputs and fewer moving parts over analogy.
6. **Validate and stop.** Check against real physics, contracts, and measured needs. Define the cheapest kill test. Stop when a coherent rebuild plus discriminating validation exists, or when only verified hard constraints remain.

## Output

Emit a first-principles rebuild:

- `claimed_constraints`: list with classification tags
- `primitives`: independently supported truths only
- `discarded_assumptions`: conventions dropped and why
- `rebuild`: solution derived only from primitives
- `binding_residuals`: real limits that remain
- `kill_test`: cheapest check that would falsify the rebuild

## Verification

- **Independence check:** every primitive must cite measurement, derivation, or primary requirement—not precedent alone.
- **No smuggled analogy:** if the rebuild still depends on "how others do it" without a primitive, strip it.
- **Constraint honesty:** verified physics/regulation must remain; do not wish them away.
- **Over-application guard:** if a standard solution is already optimal under verified constraints, do not rebuild for novelty.
- **Stop:** one decompose → classify → rebuild → kill-test pass; do not recurse without new evidence.

© tjboudreaux · MIT · tjboudreaux/cc-thinking-skills

Similar skills

All skills for Agent setup: memory, context, orchestration →

From the same set28

thinking-red-team A skill for authorized security review: it models the attacker, maps the attack surface, and keeps only findings with a reproducible exploit path and a verified fix. CX Codex +1 669installs thinking-pre-mortem Assumes a plan or launch has already failed, traces concrete causes backward, and turns them into explicit gates, owners, and verification checkpoints. CX Codex +1 660installs thinking-theory-of-constraints A skill that walks your agent through theory of constraints: find the one binding stage, exploit it, then remeasure. Useful when extra capacity does not raise output. CX Codex +1 643installs thinking-socratic Surfaces unstated assumptions and undefined terms before coding begins. Helps the agent ask decisive questions instead of guessing requirements. CX Codex +1 626installs thinking-systems A skill for problems that resist local fixes: it maps the system, finds feedback loops and delays, then ranks interventions by leverage. CX Codex +1 610installs thinking-opportunity-cost Forces your AI coding agent to identify the best forgone alternative and calculate the net value delta before committing engineering hours or budget. CX Codex +1 610installs thinking-model-router A router skill that maps domain and problem type to one thinking skill, or returns NONE. Useful when you would otherwise guess or stack frameworks. CX Codex +1 605installs thinking-steel-manning Builds the strongest opposing argument before rejecting or adopting a technical plan. Useful for architecture reviews and preventing sycophantic AI agreement. CX Codex +1 600installs

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor