THE INSTRUMENTS · THE CRAFT GATES · Nº 1339 · EDITION 1 ·Liquid error (snippets/rm-double-check-card line 27): comparison of String with 0 failed
LUND CARDS · THE INSTRUMENTSNº 1339
THE CARD AS TEXT · FOR EYES AND FOR MACHINES
THE DOUBLE CHECK CARD · the second-look seat — re-measures every claim before the work is called done
WAKES WHEN work is about to be called done and every claim in the hand-off needs re-measuring first · a person says 'double check', 'are you sure', 'verify that', 'check it again', or 'it still has a gap' · a layout has to meet an edge or fill the screen at every size, or a build has to be compared before and after
THE LOOP
list · write down every claim the hand-off makes, each with the number or the words it promises
measure · re-measure each one yourself, fresh, at every size and state it promises; write down where the evidence is
compare · the machine grades each claim CONFIRMED, CONTRADICTED or UNCHECKED, and names the gap in its own units
hold · nothing is called done while one claim is contradicted or unchecked; fix, re-measure, run again
NEVER
Confirms a claim it did not re-measure — the builder's word is not a measurement
Rounds a gap away — 5px off is 5px off unless a tolerance was written down first
Lets work be called done while any claim is contradicted or unchecked
IT MAKES claims — grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed; layout — grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed; diff — grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed; selftest — 19 checks that prove the machine.
THE LAW Nothing ships until it is proven. A craft gate checks the work against every representative case and every reading rule, records the evidence, and refuses to pass what it has not proven. The maker holds the standard; the gate holds the line.
PROOF SELFTEST 19/19 · witnessed in Claude 2026-10-21 on the studio’s own example inputs · the dated run in the box (RUN.md) · 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. LICENSE-NOTE.md rides in the packet and is the licence.
THE PUNCH · SHA-256 OF SKILL.MD · 30D728C2E7B0D6EDA62469BC
CITE THIS CARDTHE DOUBLE CHECK CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=double-check-card. Punch 30D728C2E7B0D6EDA62469BC. Retrieved 2026-10-21.
What it is.
Re-measures every claim before the work is called done. Use when work is about to be called done and every claim in the hand-off needs re-measuring first. Not for doing the building itself. For every representative case → THE EVERY REP CARD.
It carries a machine that shows its working: claims — grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed; layout — grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed; diff — grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed; selftest — 19 checks that prove the machine — 19 checks in its selftest, run in the open below. It never confirms a claim it did not re-measure — the builder's word is not a measurement; never rounds a gap away — 5px off is 5px off unless a tolerance was written down first; never lets work be called done while any claim is contradicted or unchecked.
The law, written into the card.
It is in the card itself, so the AI follows it.
THE HOUSE LAW · THE CRAFT GATESNothing ships until it is proven. A craft gate checks the work against every representative case and every reading rule, records the evidence, and refuses to pass what it has not proven. The maker holds the standard; the gate holds the line.
THE LIMIT, ON THE TINIt re-measures nothing by itself: it grades the numbers you, a browser or a test bring it, and checks a layout from measurements taken with references/measure.js. It does not open a site, and it is only as good as what was measured — when a screenshot and a CLEAR disagree, a measurement is missing.
The method.
List. Write down every claim the hand-off makes, each with the number or the words it promises.
Measure. Re-measure each one yourself, fresh, at every size and state it promises; write down where the evidence is.
Compare. The machine grades each claim CONFIRMED, CONTRADICTED or UNCHECKED, and names the gap in its own units.
Hold. Nothing is called done while one claim is contradicted or unchecked; fix, re-measure, run again.
What it does once your AI has it.
claims Grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed
layout Grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed
diff Grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed
selftest 19 checks on the studio’s own example inputs; it must print selftest 19/19
What it will never do.
It never confirms a claim it did not re-measure — the builder's word is not a measurement; never rounds a gap away — 5px off is 5px off unless a tolerance was written down first; never lets work be called done while any claim is contradicted or unchecked.
Where it hands off: Every representative case → THE EVERY REP CARD. Whether a surface reads well → THE OPTIMAL READING CARD.
It is an AI reading a text file: it can be wrong, so check what matters against your own records and the source. The holder brings their own AI; no endorsement by Anthropic or any other AI company is claimed. Names of other companies and standards are used only to say what the card works with.
What’s in the buy.
SKILL.md — the door your AI reads first: when it wakes, when it does not, the loop, the never-lines, the house law
CARD.md — the whole card in ten sections, §0 to §9 · TIN.md — what is on the tin, in nine lines
scripts/doublecheck.py — the machine, Python standard library only, offline; selftest 19/19
RUN.md — the dated run: the selftest as it printed when the card was cut, with the date and the host
references/THE-CLAIMS-FORM.csv — the claims form: claim, expected, measured, unit, tolerance, evidence
references/THE-METHOD.md — the method in plain words, in the card’s own voice
references/THE-WORKED-EXAMPLE.md — the lundr.tech desktop menu: a 5px gap and a wrong foot caught at 120 sizes, a first fix that passed the box and failed the eye, and the second fix re-measured at tolerance 0
references/measure.js — a browser-console snippet that measures where things really are at the window's real size, for the layout check
START-HERE.md and system/short-prompt.md — the doors, for any AI with or without skills
MY-LEDGER.md — the card’s memory, on your own machine · LICENSE-NOTE.md — the licence
PROOF · SELFTEST 19/19 · AUDIT 12/12 · WITNESSED IN CLAUDE 2026-10-21 ON THE STUDIO’S OWN EXAMPLE INPUTS · THE TEST PROMPTS IN FRESH CHATS OWED · EDITION 1 · PUNCH 30D728C2E7B0D6EDA62469BC
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 1339 —Liquid error (snippets/rm-double-check-card line 116): comparison of String with 0 failed
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.
AFTER YOU PAY · SIGN IN WITH THE EMAIL YOU PAID WITH · THE CARD OPENS ON MY CARDS → · CHECKED AGAINST ITS FINGERPRINT · NOTHING TO WAIT FOR
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 STUDIO’S RUN OF 2026-10-21
doublecheck.py, showing its working.
A small Python program, standard library only. Your AI runs it; you can too. The block below is the machine’s selftest as it printed when the studio ran it from inside the packet, in Claude, on 2026-10-21.
claims
grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed · claims CLAIMS.csv
layout
grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed · layout MEASURE.json --touch A:B --fill B [--reach C:B[:INSET]] [--tol PX]
diff
grades every claim CONFIRMED, CONTRADICTED or UNCHECKED, checks a layout's edges at every size, lists what changed between two folders, and holds until every claim is confirmed · diff BEFORE AFTER
selftest
19 checks that prove the machine; it must print selftest 19/19
selftest · AS PRINTED
$ python3 scripts/doublecheck.py selftest
CLEAR · 2 claims · 0 holding · wrote DOUBLE-CHECK.md
HOLD · 1 claims · 1 holding · wrote DOUBLE-CHECK.md
HOLD · 1 claims · 1 holding · wrote DOUBLE-CHECK.md
HOLD · 1 claims · 1 holding · wrote DOUBLE-CHECK.md
REFUSED: the claims file needs the columns claim, expected, measured, evidence (tolerance and unit optional)
HOLD · 2 sizes · 4 findings · wrote LAYOUT-CHECK.md
1440x900: GAP 5px · menu starts below bar's bottom edge
1440x900: PAST THE FOOT · menu runs 22px below the bottom of the screen
2560x1440: GAP 5px · menu starts below bar's bottom edge
2560x1440: SHORT OF THE FOOT · menu stops 405px above the bottom of the screen
CLEAR · 4 sizes · 0 findings · wrote LAYOUT-CHECK.md
HOLD · 1 sizes · 1 findings · wrote LAYOUT-CHECK.md
390x844: SIDEWAYS SCROLL · the page is 994px wide in a 390px window
REFUSED: say what to check: --touch A:B, --fill B and/or --reach C:B
CLEAR · 1 sizes · 0 findings · wrote LAYOUT-CHECK.md
HOLD · 1 sizes · 1 findings · wrote LAYOUT-CHECK.md
2560x1440: SHORT INSIDE · columns stops 411px above menu's foot (less its 21px padding)
1 added · 0 removed · 1 changed · wrote CHANGES.md
selftest 19/19
THE SELFTEST RUNS ON THE STUDIO’S OWN EXAMPLE INPUTS, INVENTED FOR THE TEST; NO REAL COMPANY, PERSON OR ACCOUNT IS NAMED.
QUESTIONS AT THE TABLE
Asked before the wire.
What does it need from me?
Your own files and figures, on your own disk. It starts by asking for them: write down every claim the hand-off makes, each with the number or the words it promises. If something it needs is missing, it marks it and says so — it never fills a gap with a guess.
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 machine's selftest (19/19), run from inside the packet in Claude on 2026-10-21 on the studio's own example inputs and printed below as it ran; the twelve-check audit (AUDIT.md, 12/12); MANIFEST.md with every file hashed. RUN.md in the box is that run, dated. Still to come, as the next receipt: the card's test prompts asked in fresh chats with the card installed. 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 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. It re-measures nothing by itself: it grades the numbers you, a browser or a test bring it, and checks a layout from measurements taken with references/measure.js. It does not open a site, and it is only as good as what was measured — when a screenshot and a CLEAR disagree, a measurement is missing.
What will it never do?
It never confirms a claim it did not re-measure — the builder's word is not a measurement; never rounds a gap away — 5px off is 5px off unless a tolerance was written down first; never lets work be called done while any claim is contradicted or unchecked.
