THE INTRO CARD is free — what an AI skill card is and how to use one Download →Lost Lauren, contact us

THE METHODS · THE CARD BENCH · Nº 431 · EDITION 2 · LAUNCHING SOON

LUND CARDS · THE METHODSNº 431
THE WITNESS CARD — the LSC-1 face
THE CARD AS TEXT · FOR EYES AND FOR MACHINES

THE WITNESS CARD · The proof seat: every claim beside the run that showed it — command, hashes, date, who watched.

WAKES WHEN A packet or a page says 'it works' and nobody can point at the run. · A maker wants a witness file a buyer can trust. · A claim was true last month and the input has changed since. · Someone wants to prove the output file is still the one that was recorded.

THE LOOP

  1. Record each run: command, output, input, date
  2. Check every claim against the ledger
  3. Mark the unwitnessed and the stale
  4. Write the witness file a reader sees

NEVER

  • Never records a witness with no run behind it
  • Never calls a stale claim witnessed
  • Never writes a promise where a run should be
  • Never claims a certainty it does not have

IT MAKES record — add a run: the command, the hash and last line of its output, the hash of its input; claims — check each claim against the ledger: WITNESSED (which run), UNWITNESSED, or STALE when the input changed since; write — write the witness file a reader sees: every run, and what was not witnessed; verify — does FILE still hash to what run N recorded? MATCH or CHANGED; list — print the runs.

THE LAW A witness is a run that happened: the command, the hash and last line of what it printed, the hash of what it ran on, the date, and who watched. A claim either points at a run in the ledger or is said as a claim. When the input changes, the witness is stale until the run is done again.

IN THE BOX SKILL.md · CARD.md · ▣ scripts/witness.py (machine) · ▣ scripts/evals.py (test runner) · tests/EVALS.json (9 tests) · references/ (schema + fictional demo) · START-HERE.md · system/ doors · MY-LEDGER.md · LICENSE-NOTE.md · card.svg · card.json · card.html · AUDIT.md · MANIFEST.md · WITNESS.md · SELFTEST.md.

PROOF SELFTEST 12/12 · witnessed in Claude 2026-10-14 on fictional fixtures · the test prompts in fresh chat, 9/9 · Edition 2.

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 · 871EDAC06C25BE127BB0CCCA

THE FACE, FULL SIZE → · THE MACHINE FACE → · card.json →

THE WITNESS CARD

The proof seat: every claim beside the run that showed it — command, hashes, date, who watched.

RECORD · CLAIMS · WRITE · VERIFY · LIST

98LEGENDARYWITNESSEDIN CLAUDEIN LUNDRIN CHATGPT

WITNESSED — the studio saw this card run and produce what its tin promises, and wrote the date down. What the marks mean →

THE SPEC SHEET · READ BEFORE YOU BUY · THE STANDARD EVERY CARD MEETS →

SEAL
98 of 100 · LEGENDARY · WITNESSED
EDITION
E1 · witnessed 2026-10-14
MACHINE
scripts/witness.py · 5 commands · selftest 12/12
RUNS
On your own computer. Standard library only, offline; no account, no key, no telemetry.
HOSTS
Claude · Claude Code · ChatGPT · Gemini · any AI that reads a file (START-HERE.md and system/ name each door)
LICENCE
One person, any number of their own AIs and computers · editions replace files · nothing renews · freeze, not loss
PUNCH
871EDAC06C25BE127BB0CCCA · sha-256 of SKILL.md · check any file against it →

CITE THIS CARDTHE WITNESS CARD, Edition 2. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=witness-card. Punch 871EDAC06C25BE127BB0CCCA. Retrieved 2026-10-14.

What it is.

Use when a claim about work needs its proof kept beside it: what was run, what it printed, on what input, when, witnessed by whom. Not for running the tests or for judging the writing's tone.

It carries a machine that shows its working: record — add a run: the command, the hash and last line of its output, the hash of its input; claims — check each claim against the ledger: WITNESSED (which run), UNWITNESSED, or STALE when the input changed since; write — write the witness file a reader sees: every run, and what was not witnessed; verify — does FILE still hash to what run N recorded? MATCH or CHANGED; list — print the runs — 12 checks in its selftest. It never records a witness with no run behind it; never calls a stale claim witnessed; never writes a promise where a run should be; never claims a certainty it does not have.

The law, written into the card.

It is in the card itself, so the AI follows it.

THE LAWA witness is a run that happened: the command, the hash and last line of what it printed, the hash of what it ran on, the date, and who watched. A claim either points at a run in the ledger or is said as a claim. When the input changes, the witness is stale until the run is done again.

The method.

  1. Run the thing by hand and keep its output in a file. Record it: the machine hashes the output and the input and keeps the last line as the receipt.
  2. List the claims the work makes as a sheet: the claim, the command that would show it, the input it ran on.
  3. Check the claims against the ledger: WITNESSED names the run; UNWITNESSED says why; STALE says the input moved.
  4. Write the witness file a reader sees: every run in a table, and a section for what was not witnessed, so the silence is honest too.
  5. Before shipping, verify the output file still hashes to what was recorded; CHANGED means what is there now was never witnessed.

What you bring.

  • The commands that prove the work, and their printed output kept in files.
  • The claims the work makes, written out.
  • A person to watch and be named.

Planned as a separate seat, not in this box and not yet on sale: the witness ledgers the studio keeps for every card it cuts.

What it does once your AI has it.

  • record Add a run: the command, the hash and last line of its output, the hash of its input
  • claims Check each claim against the ledger: WITNESSED (which run), UNWITNESSED, or STALE when the input changed since
  • write Write the witness file a reader sees: every run, and what was not witnessed
  • verify Does FILE still hash to what run N recorded? MATCH or CHANGED
  • list Print the runs

What it will never do.

It never records a witness with no run behind it; never calls a stale claim witnessed; never writes a promise where a run should be; never claims a certainty it does not have.

Where it hands off: Checking that the box holds what the tin says → THE TIN CARD; Reading the writing for claims that outrun their proof → THE HONEST CARD; Numbering the edition the run was on → THE EDITION CARD.

It is not an audit. 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.

What’s in the buy.

  • SKILL.md — the card your AI reads: when it wakes, the loop, the never-lines, the hand-offs.
  • CARD.md — the whole card in ten sections: face, law, method, machine, what you bring, never, tests, seal, term.
  • scripts/witness.py — the machine, Python standard library only; `selftest` proves it on its own fixtures.
  • tests/EVALS.json and scripts/evals.py — 9 test prompts and the offline runner that scores your AI's answers.
  • references/ — the data schema and FICTIONAL demo files, invented to show the machine, not real.
  • MY-LEDGER.md — your own record of what you asked and what you decided.
  • START-HERE.md, system/short-prompt.md, system/rules.md — the doors, for any AI with or without skills.
  • card.svg, card.json, card.html, AUDIT.md, MANIFEST.md, WITNESS.md, SELFTEST.md — the faces and the proof.
  • NOT IN THE BOX — any studio edition (planned, not yet on sale: the witness ledgers the studio keeps for every card it cuts). Also not in the box: any data you must bring, and advice of any kind (not an audit).

PROOF · SELFTEST 12/12 · WITNESSED IN CLAUDE 2026-10-14 ON FICTIONAL FIXTURES · THE 9 TEST PROMPTS ASKED IN A FRESH CHAT HOLDING ONLY THE CARD · 9/9 · EDITION 2 · PUNCH 871EDAC06C25BE127BB0CCCA

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 431 — 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 · NOTHING RENEWS ITSELF

OFFLINE · ONE LICENCE · PUNCH-CHECKABLE · THE STANDARD

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.

One email when it is on sale, from the studio; nothing else, ever.

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 CARD’S OWN SELFTEST.MD

witness.py, every command showing its working.

A small Python program, standard library only. Your AI runs it; you can too. The block below is pasted from the card’s own SELFTEST.md, as printed when it was run from inside the packet.

recordadd a run: the command, the hash and last line of its output, the hash of its input
claimscheck each claim against the ledger: WITNESSED (which run), UNWITNESSED, or STALE when the input changed since
writewrite the witness file a reader sees: every run, and what was not witnessed
verifydoes FILE still hash to what run N recorded? MATCH or CHANGED
listprint the runs
selftest12 checks that prove the machine; it must print selftest 12/12
THE SELFTEST, AS PRINTED
record run 1 · 2026-01-02 · `python3 scripts/machine.py selftest` → selftest 2/2 · 90a917a5a8f9
NO COMMAND: a witness line with no command behind it is refused
NO OUTPUT: <tmp> is not there; a witness with no output is refused
the machine passes its selftest                              STALE: run 1 saw a different data.csv; run it again
the evals run                                                UNWITNESSED: no run of `python3 scripts/evals.py selftest` in the ledger
it is fast                                                   UNWITNESSED: no command named
claims <tmp> · 0 witnessed · 3 not
write <tmp> · 1 runs
verify run 1 · MATCH · <tmp> still hashes to what was recorded
verify run 1 · CHANGED · <tmp> no longer matches run 1's output; what is there now was not witnessed
NO RUN 9 in <tmp>
  1  2026-01-02  python3 scripts/machine.py selftest      selftest 2/2
list <tmp> · 1 runs
ok   1  record keeps the command, the output's hash and last line, the input's hash and the date
ok   2  a witness line with no command is NO COMMAND: refused
ok   3  a witness with no output file is NO OUTPUT: refused
ok   4  claims marks the run it has WITNESSED and the ones UNWITNESSED, naming why
ok   5  a claim whose input changed since the run is STALE
ok   6  claims exits 1 when any claim is not witnessed and writes the table
ok   7  write makes the witness file with every run and a not-witnessed section
ok   8  verify says MATCH while the output file is what was recorded
ok   9  a changed output is CHANGED: what is there now was not witnessed
ok   10  a run number not in the ledger is NO RUN
ok   11  a file that is not a ledger is refused as bad input
ok   12  list prints the runs
selftest 12/12

THE RUN IS PASTED FROM THE CARD’S OWN SELFTEST.MD AS PRINTED. ITS FIXTURES ARE FICTIONAL, INVENTED FOR THE TEST; NO PRIVATE PERSON IS NAMED.

QUESTIONS AT THE TABLE

Asked before the wire.

What does it need from me?

The commands that prove the work, and their printed output kept in files. The claims the work makes, written out. A person to watch and be named. If something it needs is missing, it says what is missing and stops there.

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 (12/12), run from inside the packet in Claude on 2026-10-14 on fictional fixtures and pasted into SELFTEST.md as printed; the twelve-check audit (AUDIT.md). Owed, and said so in WITNESS.md: Any run on real data. Every demo file is fictional. 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. LICENSE-NOTE.md rides in the packet and is the licence. It is not an audit.

What will it never do?

It never records a witness with no run behind it; never calls a stale claim witnessed; never writes a promise where a run should be; never claims a certainty it does not have.

← THE METHODS · THE FLOOR →

AT YOUR ELBOW, NEVER THE BOSS · NOT AN AUDIT · NOT A PAYMENT CARD · NO REFUNDS · © 2026 LUND STUDIO LLC

PAY BYAmerican Express Apple Pay Diners ClubDiscoverGoogle Pay Mastercard PayPalShop Pay Visa
CART · 0