THE CLARIFY CARD · ask or proceed, and the one question to ask
WAKES WHEN a request is unclear and the agent must choose between asking a question and making a guess
THE LOOP
clarify.py score · open, vague, cost
ASK → the one question; nothing else
PROCEED → assume · the list, said aloud
irreversible → always shown first
the answer turns the work with one word
NEVER
Ask a list of questions
Guess at anything that cannot be undone
Proceed without saying the assumptions
IT MAKES score — open dimensions, cost, the decision, the one question; questions — every candidate question, ranked; assume — the assumptions to state when proceeding.
THE STANDARD one question, the best one; irreversible asks first; a pronoun with nothing before it is the first question; cost sets the bar; assumptions are said aloud; the machine weighs; the agent asks.
PROOF SELFTEST 10/10 · witnessed in Claude 2026-10-09 · Edition 1.
THE SEAL 98 of 100 · LEGENDARY · WITNESSED · Trigger 5 · Machinery 5 · Law 5 · Portability 5 · Proof 5
IT CAN BE WRONG an AI reading a text file: it can be wrong and can invent a number, a rule, a name or a date; check what matters against your own records and the source; the risk is yours. THE-CLAUSES.md rides in the packet and is part of the licence.
THE PUNCH · SHA-256 OF SKILL.MD · 03EFE3279A137994D8971C01
CITE THIS CARDTHE CLARIFY CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=clarify-card. Punch 03EFE3279A137994D8971C01. Retrieved 2026-10-10.
What it is.
Decides whether a request should be asked about or acted on: reads it for the open dimensions (the thing to make, who for, how big, what form, by when), vague words and pronouns with no antecedent, weighs them against the cost of a wrong guess (irreversible actions, large work), and returns ASK with the single best question, ranked, or PROCEED with the assumptions to state aloud.
It carries a machine that shows its working: score — open dimensions, cost, the decision, the one question; questions — every candidate question, ranked; assume — the assumptions to state when proceeding — 10 checks in its selftest, every command on the record. It will never ask a list of questions; never guess at anything that cannot be undone; never proceed without saying the assumptions.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
ONE QUESTION, THE BEST ONEA list of questions is a form, and people do not fill in forms to get help.
IRREVERSIBLE ASKS FIRSTIrreversible asks first, whatever else is clear: delete, send, publish, pay are shown before they are done.
A PRONOUN WITH NOTHING BEFORE IT IS THE FIRST QUESTION"Fix it" costs more than any other two words.
COST SETS THE BARHigh cost asks at the first open thing; mid at three; low proceeds and says its assumptions.
ASSUMPTIONS ARE SAID ALOUDAssumptions are said aloud, so the person can turn the work with one word instead of a conversation.
THE MACHINE WEIGHS; THE AGENT ASKSThe question is written in the person's own terms, not the dimension's name.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/ASK-OR-PROCEED.md
THE MACHINEclarify.py · three commandsscore — open dimensions, cost, the decision, the one question; questions — every candidate question, ranked; assume — the assumptions to state when proceeding. Every command prints its working and refuses a bad input; 10 checks in its selftest, every command on the record.
What it does once your AI has it.
score Open dimensions, cost, the decision, the one question
questions Every candidate question, ranked
assume The assumptions to state when proceeding
On the record Three requests weighed: "fix it and clean up the stuff" asked, with the antecedent as the one question and the rest ranked behind it; a clear one-page summary request proceeded with its assumptions written to say aloud; "delete all the old drafts" asked first about the deletion, whatever else was clear.
What it will never do.
It will never ask a list of questions; never guess at anything that cannot be undone; never proceed without saying the assumptions.
It reads words, not context: a request that is clear from the last ten turns can still score as open, and the agent brings the context. The dimension and vague-word lists are conventions in the file. English only. It is an AI reading a text file: it can be wrong, so check what matters against your own records and the source.
What’s in the buy.
CARD.md — the whole card: who it is, its laws, when it says no, what it carries · SKILL.md — the door your AI reads first · START-HERE.md — the first five minutes, for you
ASK-OR-PROCEED.md — the guide in references/, in the card's own voice
scripts/clarify.py — the machine: score · questions · assume, standard library only, selftest 10/10
WITNESS.md + SELFTEST.md — every command on the record, as it printed · MANIFEST.md, every file hashed · AUDIT.md, the twelve checks
MY-LEDGER.md — the card's memory, on your own machine · TIN.md · LICENSE-NOTE.md · THE-CLAUSES.md · HOSTS.md
card.svg + card-machine.svg + card.json · card.html, the card as one offline file · the doors for ChatGPT, Gemini, Cursor, Claude Code and MCP hosts (rules/, system/, mcp/)
PROOF · SELFTEST 10/10 · 3 COMMANDS ON THE RECORD · WITNESSED IN CLAUDE 2026-10-09 · EDITION 1 · PUNCH 03EFE3279A137994D8971C01
ON EVERY DOOR · A CARD FOR YOUR AI, A FILE FOR YOUR COMPUTER
In Claude: turn on code execution under Settings › Capabilities, then Customize › Skills › + › Create skill, and upload the zip as it is — START-HERE.md in the box walks you through it, and tells you how to keep the ledger. In ChatGPT, Gemini and the rest: a Project, a GPT or a Gem with the card’s files attached and the short door in the box pasted as its instructions. The words run anywhere an AI reads Markdown; the machines run wherever the host runs Python. Witnessed in Claude only. The host marks →
Take it.
Nº 905 PROPOSED — LAUNCHING SOON
Sold as is under the term printed. A card instructs an AI you operate; results depend on your model, your files and your judgement. Not medical, legal, tax or financial advice, never the person’s boss, and not a licence to practise. The terms →
One payment. The file is yours to keep and keeps working at its edition; new editions issued in the year after you buy reach you by email as replacement files. Tax is added at checkout where it applies. Nothing auto-renews.
A TERM, NOT A SUBSCRIPTION: NOTHING RENEWS ITSELF AND NOBODY IS CHARGED AGAIN. WHEN A TERM ENDS THE FILES STAY ON YOUR COMPUTER AND STAY READABLE — FREEZE, NOT LOSS.
THE MACHINE · IN THE OPEN · FROM THE WITNESSED RUN OF 2026-10-09
clarify.py, every command showing its working.
A small Python program, standard library only. Your AI runs it; you can too. Every block below is pasted from the card’s own WITNESS.md, as printed.
score
open dimensions, cost, the decision, the one question
questions
every candidate question, ranked
assume
the assumptions to state when proceeding
selftest
10 checks that prove the machine; it must print 10/10
score · OPEN DIMENSIONS, COST, THE DECISION, THE ONE QUESTION
$ python3 scripts/clarify.py score r1.txt
ASK · cost low · open: deliverable, audience, scope, format, deadline · vague: clean up, etc, stuff · opens on a pronoun
the question: Which thing exactly — the file, the message or the version from earlier? Name it and I will not guess.
questions · EVERY CANDIDATE QUESTION, RANKED
$ python3 scripts/clarify.py questions r1.txt
1. [antecedent] Which thing exactly — the file, the message or the version from earlier? Name it and I will not guess.
2. [deliverable] What should come back: a file, a list, a number, a draft? In what form?
3. [scope] How many, and which ones: the whole set or a sample, this file or every file?
4. [format] In what format, and where should it live when done?
5. [audience] Who is it for, and what do they already know?
6. [deadline] By when, and is a rough version sooner better than a finished one later?
7. [vague] What would "clean up" look like when it is done right?
8. [vague] What would "etc" look like when it is done right?
score · OPEN DIMENSIONS, COST, THE DECISION, THE ONE QUESTION
$ python3 scripts/clarify.py score r2.txt
PROCEED · cost mid · open: none · vague: none
proceed, and say the assumptions aloud (assume)
assume · THE ASSUMPTIONS TO STATE WHEN PROCEEDING
$ python3 scripts/clarify.py assume r2.txt
# Assumptions, said aloud
Request: Write a one-page markdown summary of attached-notes.md for the team by Friday; three bullets per section, no more than 300 words.
If any line is wrong, say so and the work turns on the spot.
score · OPEN DIMENSIONS, COST, THE DECISION, THE ONE QUESTION
$ python3 scripts/clarify.py score r3.txt
ASK · cost high · open: deliverable, audience, format, deadline · vague: none · irreversible: delete
the question: This would delete — do you want that done now, or shown first?
THE NUMBERS, NAMES AND DATES IN THESE RUNS ARE THE STUDIO’S OWN EXAMPLES, INVENTED FOR THE RECORD; NO REAL COMPANY, PERSON OR PROPERTY IS NAMED.
QUESTIONS AT THE TABLE
Asked before the wire.
What does it need from me?
Clarify.py score · open, vague, cost. It asks first for the request as it arrived, and whether a wrong guess could be undone.
Does it send anything anywhere?
No. The machine is a standard-library Python program that runs on your own computer and sends nothing anywhere; the words run wherever an AI reads Markdown. No server, no account, no key.
What is on the record?
The self-test (10/10) and every command run on real example inputs, pasted into WITNESS.md as they printed; the punch on the face is the SHA-256 of SKILL.md, so a downloaded copy can be checked against this room.
Can it be wrong?
Yes — it is an AI reading a text file. It reads words, not context: a request that is clear from the last ten turns can still score as open, and the agent brings the context. The dimension and vague-word lists are conventions in the file. English only.
What will it never do?
It will never ask a list of questions; never guess at anything that cannot be undone; never proceed without saying the assumptions.
