First principles thinking

Skill risadams/ink-and-agency/skills/research-analysis/first-principles-thinking

A dual-host skills plugin for Claude Code and OpenAI Codex with a self-evolve loop that learns from every invocation.

Install

npx -y skills add risadams/ink-and-agency --skill first-principles-thinking

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when the user wants to challenge assumptions, break down a complex problem from scratch, or approach something with first principles reasoning. Triggers on: 'first principles', 'challenge assumptions', 'why do we do it this way', 'rethink', 'from scratch', 'fundamental truths'.

SKILL.md

3.2 KB, as published. Nobody here has run it

First Principles Thinking

You strip a problem back to what is actually known and rebuild from there. The value is in finding the assumption nobody noticed they were making.

Separate what is verified from what is inherited

List the claims the current approach rests on and mark each: physical or mathematical constraint, verified fact, industry convention, or assumption never tested. Most of what feels like a constraint is the fourth category. "We can't because X" is the sentence to interrogate, and the interrogation is simply: how do we know X?

Attack analogy-based reasoning specifically

"Everyone in this industry does it this way" explains what is, not what must be. Conventions encode the constraints of the moment they were formed, and those constraints frequently no longer hold. Ask what changed since this practice was established.

Rebuild without importing the old shape

Having decomposed, construct the solution from the genuine constraints alone. The failure mode is decomposing and then reassembling the existing answer with new justification — check whether your conclusion could have been reached without the analysis.

Quantify where the reasoning turns on a number

Order-of-magnitude estimates settle arguments that qualitative reasoning cannot. If a claim rests on a cost, a rate, or a limit, put a number on it — the number frequently makes the debate disappear in one direction or the other.

Respect the constraints that are real

First-principles reasoning that ignores regulation, physics, existing contracts, or the organization's actual capacity produces confident nonsense. The point is to distinguish real constraints from imagined ones, not to deny that any exist.

Know the cost

This is expensive thinking and most decisions do not warrant it. Reserve it for decisions that are consequential, expensive to reverse, or where the conventional answer keeps producing bad outcomes.

Reporting

Present the assumptions found and their status, the constraints that survived scrutiny, the reconstructed approach, and specifically what changed versus the conventional answer.

Host portability: tool names in this skill follow Claude Code conventions; on other hosts (Codex, opencode) map them by intent — see PORTABILITY.md.

Self-Evolve Loop

Journal: ~/.ink-and-agency/learnings/first-principles-thinking.md (workspace-local .ink-and-agency/learnings/first-principles-thinking.md where the sandbox confines writes). Read it first, append what the run taught last — SELF-EVOLVE.md.