THE FORGET CARD · what an assistant saved and should not have
WAKES WHEN someone asks what their assistant remembers or wants a memory file reviewed and cleaned
THE LOOP
forget.py scan · every flagged line, why
report · the counts by reason
list · the forget list; the person reads
forget --lines · a copy; the original kept
put the copy in place only on their yes
NEVER
Remove a line the person has not approved
Write the original in place
Keep an inference as if it were a fact
IT MAKES scan — every flagged line with its reason; list — the forget list for those reasons, ready to approve; forget — the file without those lines; the original untouched; report — counts by reason.
THE STANDARD consent is the design; the original is never written; a memory is what was said; someone else's claim is not the person's fact; numbers are checked, not guessed; the lexicons are a floor.
PROOF SELFTEST 12/12 · 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 · 14E78CB50B444D05C32B442B
CITE THIS CARDTHE FORGET CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=forget-card. Punch 14E78CB50B444D05C32B442B. Retrieved 2026-10-10.
What it is.
Reviews what an assistant has saved about a person for the lines that should not be there — health, finances, protected attributes, legal history, card and ID numbers (checked with Luhn), inferences nobody stated, hearsay, dated lines that still say "currently" — names each with its reason, builds the forget list for the person to approve, and writes a mended copy with the original untouched.
It carries a machine that shows its working: scan — every flagged line with its reason; list — the forget list for those reasons, ready to approve; forget — the file without those lines; the original untouched; report — counts by reason — 12 checks in its selftest, every command on the record. It will never remove a line the person has not approved; never write the original in place; never keep an inference as if it were a fact.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
CONSENT IS THE DESIGNThe machine proposes; the person approves line numbers; only then is a copy written.
THE ORIGINAL IS NEVER WRITTENA mended copy beside it, always; the person puts it in place.
A MEMORY IS WHAT WAS SAIDA memory is what was said, never what was concluded: "seems", "probably", "the kind of person" are flagged as inference.
SOMEONE ELSE'S CLAIM IS NOT THE PERSON'S FACTHearsay is flagged and goes.
NUMBERS ARE CHECKED, NOT GUESSEDA card number is one that passes Luhn; a random run of digits is left alone.
THE LEXICONS ARE A FLOORThey catch the plain words; the person reads for the rest.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/WHAT-SHOULD-NOT-BE-REMEMBERED.md
THE MACHINEforget.py · four commandsscan — every flagged line with its reason; list — the forget list for those reasons, ready to approve; forget — the file without those lines; the original untouched; report — counts by reason. Every command prints its working and refuses a bad input; 12 checks in its selftest, every command on the record.
What it does once your AI has it.
scan Every flagged line with its reason
list The forget list for those reasons, ready to approve
forget The file without those lines; the original untouched
report Counts by reason
On the record An example profile scanned: the bakery, the bullet-point preference and the flour left alone; a diagnosis, a salary, a card number that passes Luhn, a guess about the person and a colleague's hearsay all named with their reasons, and a dated "currently" past its term read stale; the forget list approved line by line and a mended copy written beside the untouched original.
What it will never do.
It will never remove a line the person has not approved; never write the original in place; never keep an inference as if it were a fact.
Keyword lexicons miss the oblique and flag the innocent ("white background" reads as protected); the person judges every flag. English only. It reads the files it is given; it does not reach into any assistant's own memory store. 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
WHAT-SHOULD-NOT-BE-REMEMBERED.md — the guide in references/, in the card's own voice
scripts/forget.py — the machine: scan · list · forget · report, standard library only, selftest 12/12
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 12/12 · 4 COMMANDS ON THE RECORD · WITNESSED IN CLAUDE 2026-10-09 · EDITION 1 · PUNCH 14E78CB50B444D05C32B442B
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 902 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
forget.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.
scan
every flagged line with its reason
list
the forget list for those reasons, ready to approve
forget
the file without those lines; the original untouched
report
counts by reason
selftest
12 checks that prove the machine; it must print 12/12
scan · EVERY FLAGGED LINE WITH ITS REASON
$ python3 scripts/forget.py scan memory
memory/profile.md
6 · stale · dated 2024-01-15 and still says "currently"
- [stated] 2024-01-15: currently training for a half marathon in March
7 · sensitive · health
- [stated] was diagnosed with type 2 diabetes last spring and tracks carbs
8 · sensitive · finances
- [stated] salary is 54,000 a year after the raise
9 · identifier · card number
- card on file 4111 1111 1111 1111 for the subscription
10 · inference · seems
- seems anxious about money and probably avoids conflict
11 · hearsay · according to
- according to a colleague, is leaving the bakery
12 · sensitive · protected
- [stated] votes for the green party every election
7 lines flagged of 14 · hearsay 1, identifier 1, inference 1, sensitive 3, stale 1
list · THE FORGET LIST FOR THOSE REASONS, READY TO APPROVE
$ python3 scripts/forget.py list memory/profile.md --reasons sensitive,identifier,inference,hearsay
forget list for sensitive, identifier, inference, hearsay · 6 lines · approve, then: forget memory/profile.md --lines 7,8,9,10,11,12
7 · sensitive · health · - [stated] was diagnosed with type 2 diabetes last spring and tracks carbs
8 · sensitive · finances · - [stated] salary is 54,000 a year after the raise
9 · identifier · card number · - card on file 4111 1111 1111 1111 for the subscription
10 · inference · seems · - seems anxious about money and probably avoids conflict
11 · hearsay · according to · - according to a colleague, is leaving the bakery
12 · sensitive · protected · - [stated] votes for the green party every election
(the person reads the list and approves lines 7, 8, 9, 10 and 11)
$ (the person reads the list and approves lines 7, 8, 9, 10 and 11)
forget · THE FILE WITHOUT THOSE LINES; THE ORIGINAL UNTOUCHED
$ python3 scripts/forget.py forget memory/profile.md --lines 7,8,9,10,11 --out memory/profile.mended.md
wrote memory/profile.mended.md without 5 lines; memory/profile.md is untouched
cat memory/profile.mended.md
$ cat memory/profile.mended.md
---
name: profile
---
- [stated] runs a small bakery in Leeds; opens at 6 and likes the early shift
- [stated] prefers answers as bullet points, short
- [stated] 2024-01-15: currently training for a half marathon in March
- [stated] votes for the green party every election
- [stated] favourite flour is a stoneground rye from a mill near Hebden Bridge
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?
Forget.py scan · every flagged line, why. It asks first for the memory or profile files, and the person's own eye on the forget list before anything is removed.
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 (12/12) 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. Keyword lexicons miss the oblique and flag the innocent ("white background" reads as protected); the person judges every flag. English only. It reads the files it is given; it does not reach into any assistant's own memory store.
What will it never do?
It will never remove a line the person has not approved; never write the original in place; never keep an inference as if it were a fact.
