THE BRIEF CARD · the hand-off brief, checked before it goes
WAKES WHEN someone is about to hand work to another agent or asks whether a brief is good enough to send
THE LOOP
brief.py template · the seven parts
fill it from the work, not from memory
check · score, missing parts, vague words
mend what is named; check again
cut · the one-screen brief; send that
NEVER
Send a brief with no done-when test
Name "the relevant files" instead of paths
Let a pronoun stand with nothing before it
IT MAKES template — the seven parts, with what each must hold; check — the score out of 100, what is missing, what is vague; cut — the one-screen brief, the seven parts in order.
THE STANDARD seven parts, every time; done-when is a test; paths, not hints; out is written down; vague words are named; the score is the shape.
PROOF SELFTEST 11/11 · 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 · 389064E1C009D2FA977DDA18
CITE THIS CARDTHE BRIEF CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=brief-card. Punch 389064E1C009D2FA977DDA18. Retrieved 2026-10-10.
What it is.
Writes and checks the brief one agent hands another before delegating work: the goal in one sentence, the test that says it is done, what was ruled out, what to read first, the scope in and out, the shape of the answer and what must not be touched — scored out of 100 with every missing or vague part named, and cut to one screen.
It carries a machine that shows its working: template — the seven parts, with what each must hold; check — the score out of 100, what is missing, what is vague; cut — the one-screen brief, the seven parts in order — 11 checks in its selftest, every command on the record. It will never send a brief with no done-when test; never name "the relevant files" instead of paths; never let a pronoun stand with nothing before it.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
SEVEN PARTS, EVERY TIMEA brief missing a part is not shorter; it is a guess the receiver makes for the sender.
DONE-WHEN IS A TESTA number, a file, a command that passes. "When it works" is not a test.
PATHS, NOT HINTSREAD FIRST names files and links; "the relevant docs" sends the receiver searching.
OUT IS WRITTEN DOWNThe scope line says what is out; that line stops the receiver rebuilding the world.
VAGUE WORDS ARE NAMED"Clean up", "as needed", "etc" are flagged and rewritten before sending.
THE SCORE IS THE SHAPEA brief can score 100 and ask for the wrong thing; the sender owns the wisdom, the card owns the form.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/THE-SEVEN-PARTS.md
THE MACHINEbrief.py · three commandstemplate — the seven parts, with what each must hold; check — the score out of 100, what is missing, what is vague; cut — the one-screen brief, the seven parts in order. Every command prints its working and refuses a bad input; 11 checks in its selftest, every command on the record.
What it does once your AI has it.
template The seven parts, with what each must hold
check The score out of 100, what is missing, what is vague
cut The one-screen brief, the seven parts in order
On the record The template written; a real brief for the shelf-rail job checked and sent at 93/100 with nothing missing; a one-line "look into the shelf thing" checked and refused at 30/100 with all seven parts missing; the real brief cut to one screen in the seven-part order.
What it will never do.
It will never send a brief with no done-when test; never name "the relevant files" instead of paths; never let a pronoun stand with nothing before it.
It scores the shape of a brief, not its wisdom: a complete brief for the wrong task scores well. Its vague-word list is a convention, edited in the file. It reads headings in English (GOAL, DONE WHEN…) and their listed aliases. 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
THE-SEVEN-PARTS.md — the guide in references/, in the card's own voice
scripts/brief.py — the machine: template · check · cut, standard library only, selftest 11/11
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 11/11 · 3 COMMANDS ON THE RECORD · WITNESSED IN CLAUDE 2026-10-09 · EDITION 1 · PUNCH 389064E1C009D2FA977DDA18
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 897 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
brief.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.
template
the seven parts, with what each must hold
check
the score out of 100, what is missing, what is vague
cut
the one-screen brief, the seven parts in order
selftest
11 checks that prove the machine; it must print 11/11
template · THE SEVEN PARTS, WITH WHAT EACH MUST HOLD
$ python3 scripts/brief.py template --out brief.md
wrote brief.md
(the sender fills the seven parts for a real task: the shelf rail recounted from the rows)
$ (the sender fills the seven parts for a real task: the shelf rail recounted from the rows)
check · THE SCORE OUT OF 100, WHAT IS MISSING, WHAT IS VAGUE
$ python3 scripts/brief.py check brief.md
BRIEF · 100/100 · SEND
GOAL · 10/10
DONE WHEN · 10/10
RULED OUT · 10/10
READ FIRST · 10/10
SCOPE · 10/10
ANSWER SHAPE · 10/10
DO NOT TOUCH · 10/10
check · THE SCORE OUT OF 100, WHAT IS MISSING, WHAT IS VAGUE
$ python3 scripts/brief.py check vague.md
BRIEF · 20/100 · NOT A BRIEF YET
MISSING · GOAL — one sentence, a verb and the thing that will exist when the work is done
MISSING · DONE WHEN — the test the receiver can run to know it is finished: a number, a file, a command that passes
MISSING · RULED OUT — what has already been tried or decided against, so it is not tried again
MISSING · READ FIRST — the files, docs or links to read before starting, by path or title
MISSING · SCOPE — what is in and, in its own line, what is out
MISSING · ANSWER SHAPE — the form the answer comes back in: a file, a table, three bullets, a yes or no with a reason
MISSING · DO NOT TOUCH — what must not change, be deleted, sent or spent without asking
missing: GOAL, DONE WHEN, RULED OUT, READ FIRST, SCOPE, ANSWER SHAPE, DO NOT TOUCH
cut · THE ONE-SCREEN BRIEF, THE SEVEN PARTS IN ORDER
$ cat short.md
# BRIEF · the shelf rail recounted
**GOAL** — Write a script that counts the shelf's rows by house and rewrites the rail's numbers from the count.
**DONE WHEN** — `python3 shelfside.py theme` prints 735 rows and every house count on the rail equals the rows that carry its class; 0 houses without a filter.
**RULED OUT** — Typing the counts by hand (that is how they went stale). Regenerating the whole shelf page (the rows are right; only the rail is wrong).
**READ FIRST** — snippets/rm-shelf.liquid (the rail and the CSS rules), snippets/shelf-rows-1.liquid to shelf-rows-4.liquid, assets/floor-index.json.
**SCOPE** — In: the rail's counts, the house filters and their CSS rules. Out: the rows themselves, the sort order, the store's till flags.
**ANSWER SHAPE** — The script as a file, and a one-line report: rows, houses, and the three counts that changed.
**DO NOT TOUCH** — Do not delete any row; do not change the row markup; do not push or publish anything.
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?
Brief.py template · the seven parts. It asks first for the task as the sender understands it, the files the receiver must read, and what must not be touched.
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 (11/11) 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 scores the shape of a brief, not its wisdom: a complete brief for the wrong task scores well. Its vague-word list is a convention, edited in the file. It reads headings in English (GOAL, DONE WHEN…) and their listed aliases.
What will it never do?
It will never send a brief with no done-when test; never name "the relevant files" instead of paths; never let a pronoun stand with nothing before it.
