THE CUT LIST · what every turn pays for; what no run used
WAKES WHEN someone asks what the always-on files cost every turn, or what could be cut from them
THE LOOP
budget.py measure · tokens, file by file
unused · against a month of transcripts
read the cut list; a rare rule may earn it
trim --out · a copy, never the original
run on the copy a week; then replace
NEVER
Delete a section on the count alone
Call the estimate a tokenizer: it is a count
Trim the original in place
IT MAKES measure — tokens per file and per section, the per-turn total; unused — the sections whose words never appear in any transcript; trim — the file without those sections, and the saving.
THE STANDARD every turn pays for every line; a count, not a tokenizer; unused is where to look, not what to delete; trim to a copy; stop words are not evidence; say the saving in tokens a turn.
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 · 58DC7905B5475DA9002D4242
CITE THIS CARDTHE CUT LIST, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=cut-list-card. Punch 58DC7905B5475DA9002D4242. Retrieved 2026-10-10.
What it is.
Measures what the always-loaded files cost on every turn — a CLAUDE.md, the rules files, a skill set — in tokens per file and per section, finds the sections whose own distinctive words never appear in any transcript, and writes a trimmed copy with the tokens saved.
It carries a machine that shows its working: measure — tokens per file and per section, the per-turn total; unused — the sections whose words never appear in any transcript; trim — the file without those sections, and the saving — 10 checks in its selftest, every command on the record. It will never delete a section on the count alone; never call the estimate a tokenizer: it is a count; never trim the original in place.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
EVERY TURN PAYS FOR EVERY LINEWhat is loaded always is read always; the cost is the file times the turns.
A COUNT, NOT A TOKENIZERWords × 1.3 and characters ÷ 4, the higher kept — the same error on every file, so the order holds.
UNUSED IS WHERE TO LOOK, NOT WHAT TO DELETEA section no transcript touched is a question for the person who wrote it.
TRIM TO A COPYThe original stays; the trimmed file runs for a week before it replaces anything.
STOP WORDS ARE NOT EVIDENCEA section is used when its distinctive words appear, not when "the" does.
SAY THE SAVING IN TOKENS A TURNSay the saving in tokens a turn, because that is the unit the bill is in.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/WHAT-A-TURN-COSTS.md
THE MACHINEbudget.py · three commandsmeasure — tokens per file and per section, the per-turn total; unused — the sections whose words never appear in any transcript; trim — the file without those sections, and the saving. 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.
measure Tokens per file and per section, the per-turn total
unused The sections whose words never appear in any transcript
trim The file without those sections, and the saving
On the record A rules file measured section by section; against two transcripts the build, deploy and style sections were used and the legacy ftp mirror section never was; trimmed to a copy with the tokens saved per turn, and the copy measured again.
What it will never do.
It will never delete a section on the count alone; never call the estimate a tokenizer: it is a count; never trim the original in place.
The token figure is an estimate within about 15% for English and code. "Unused" is a word-match against transcripts: a rule followed silently, never quoted, can read as unused. Sections are read from markdown headings; a file with no headings is one section. 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-A-TURN-COSTS.md — the guide in references/, in the card's own voice
scripts/budget.py — the machine: measure · unused · trim, 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 58DC7905B5475DA9002D4242
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 899 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
budget.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.
measure
tokens per file and per section, the per-turn total
unused
the sections whose words never appear in any transcript
trim
the file without those sections, and the saving
selftest
10 checks that prove the machine; it must print 10/10
measure · TOKENS PER FILE AND PER SECTION, THE PER-TURN TOTAL
unused · THE SECTIONS WHOSE WORDS NEVER APPEAR IN ANY TRANSCRIPT
$ python3 scripts/budget.py unused CLAUDE.md transcripts
USED 23 tok Build hits 2/6 · build, error
USED 25 tok Deploy hits 4/5 · deploy, green, production, staging
UNUSED 27 tok Legacy ftp mirror hits 0/8
USED 21 tok Style hits 4/8 · indentation, quotes, single, two-space
cut list: 1 section, ~27 tokens a turn
trim · THE FILE WITHOUT THOSE SECTIONS, AND THE SAVING
$ python3 scripts/budget.py trim CLAUDE.md --drop Legacy ftp mirror --out CLAUDE.trimmed.md
wrote CLAUDE.trimmed.md · ~33 tokens saved a turn
measure · TOKENS PER FILE AND PER SECTION, THE PER-TURN TOTAL
$ python3 scripts/budget.py measure CLAUDE.trimmed.md
80 CLAUDE.trimmed.md
23 Build
25 Deploy
21 Style
every turn: ~80 tokens across 1 file
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?
Budget.py measure · tokens, file by file. It asks first for the files loaded every turn, and the transcripts of the sessions that ran with them.
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. The token figure is an estimate within about 15% for English and code. "Unused" is a word-match against transcripts: a rule followed silently, never quoted, can read as unused. Sections are read from markdown headings; a file with no headings is one section.
What will it never do?
It will never delete a section on the count alone; never call the estimate a tokenizer: it is a count; never trim the original in place.
