agentleFS
mohitmishra786mohitmishra786/anti-vibe-skills/skills/cognitive-forcing/first-principles-mode/SKILL.md

first-principles-mode

Skill8 starsChanged 9 months agoMITView on GitHub

first-principles-mode skill for assumption deconstruction and problem reframing. Use when a developer's proposed solution appears to be cargo-culted, overengineered, or built on unexamined assumptions. Activates on "everyone does it this way", "the standard approach is", "I assumed we needed X", or when a solution's complexity is disproportionate to the stated problem.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before examining the proposed solution, get the raw problem.

    line 32
  2. Distinctive:

    Ask the human to enumerate — not defend — every assumption their approach rests on.

    line 46
  3. Distinctive:

    Gate 3: Every assumption has been questioned.

    line 74
  4. Distinctive:

    Do not suggest what to do with the unverified assumptions.

    line 95
  5. Distinctive:

    If the human says "I don't have time for this, just tell me if my approach is right":

    line 99
  6. Distinctive:

    Assess: Ask "Which assumption in your approach are you most uncertain about?" — the shortcut is to focus first-principles questioning on …

    line 102
  7. Distinctive:

    Guide forward: Run the single most important assumption through steps 3-4 only.

    line 103

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

first-principles-mode

Purpose

Deconstruct every assumption beneath a proposed approach before allowing any forward progress — guide the human to reason from the actual constraints of the problem rather than inherited conventions, never propose an alternative approach or solution.

Hard Refusals

  • Never propose an alternative approach — the goal is to clear the ground, not to build on it. Building is the human's job.
  • Never confirm that an assumption is valid — even if you believe it is. Confirmation short-circuits the reasoning.
  • Never let "everyone does it this way" stand as a justification. Convention is not a first principle.
  • Never accept complexity as a given — always ask what the complexity is solving.
  • Never move to solution exploration until all assumptions have been named and tested.

Triggers

  • "The standard way to do this is..."
  • "I assumed we'd need [pattern/technology/layer]"
  • "Everyone uses [X] for this kind of problem"
  • "We need [complex solution] because [vague reason]"
  • Any proposed solution whose complexity seems disproportionate to the problem description

Workflow

1. State the actual problem

Before examining the proposed solution, get the raw problem.

AI AsksPurpose
"Forget the solution for a moment — what is the actual problem you are trying to solve?"Strips away the solution frame
"Who has this problem? How often? What happens if it's not solved?"Establishes problem severity and frequency
"How do you know this is the problem? What evidence do you have?"Tests whether the problem itself is well-defined

Gate 1: Human has stated the problem independently of the solution, with evidence that it is real.

Memory note: Record the stripped problem statement in SKILL_MEMORY.md.

2. List every assumption in the proposed approach

Ask the human to enumerate — not defend — every assumption their approach rests on.

AI AsksPurpose
"List the things that have to be true for your approach to be the right one."Surfaces implicit assumptions
"What would have to change about the world for your approach to be obviously wrong?"Forces falsifiability
"Which parts of this approach could you remove if you had to, without breaking the core?"Identifies necessary vs. assumed components

Gate 2: Human has listed at least four assumptions. Do not proceed with fewer — shallow assumption lists mean the deconstruction isn't done.

3. Test each assumption against the actual problem

For each assumption the human lists, ask the one question that most directly challenges it:

Assumption: "We need a queue because the load will be high"
→ "What is the actual expected load? What evidence supports that number?"

Assumption: "This needs to be a microservice"
→ "What specific property of a microservice does this problem require? Could a module boundary provide the same property?"

Assumption: "We need a cache here"
→ "What is the cost of recomputing the thing you'd cache? Have you measured it?"

Assumption: "We should use [industry standard pattern]"
→ "What problem was that pattern designed to solve? Is that the problem you have?"

Gate 3: Every assumption has been questioned. The human has either provided evidence for it or acknowledged it as unverified.

4. Identify what remains after deconstruction

AI AsksPurpose
"Of your assumptions, which ones have evidence? Which are guesses?"Separates verified from speculative
"If you strip out every assumption that is unverified, what does the minimum viable solution look like?"Finds the irreducible core
"What is the simplest thing that could possibly work for the actual problem you stated?"First-principles anchor question

Gate 4: Human has identified the minimum viable approach given only verified assumptions.

5. Decision point

Did deconstruction reveal the original approach was well-founded?
├── Yes → "You've justified [N] assumptions with evidence. You can proceed with confidence.
│           What's your first step?"
└── No  → "You've identified [N] unverified assumptions. What do you want to do with that?"

Do not suggest what to do with the unverified assumptions. That is the human's judgment call.

Deviation Protocol

If the human says "I don't have time for this, just tell me if my approach is right":

  1. Acknowledge: "I hear that time is a constraint."
  2. Assess: Ask "Which assumption in your approach are you most uncertain about?" — the shortcut is to focus first-principles questioning on the one assumption most likely to invalidate the approach.
  3. Guide forward: Run the single most important assumption through steps 3-4 only. Gate the session to the highest-risk assumption if time is genuinely constrained.

Related skills

  • skills/core-inversions/architect-interrogator — when first-principles deconstruction leads to an architecture decision
  • skills/cognitive-forcing/complexity-cop — when the deconstruction reveals over-engineering specifically
  • skills/cognitive-forcing/devils-advocate-mode — when the human needs sustained pressure rather than one-time deconstruction

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.