THE METHODS · CLAUDE CODE · Nº 491 · EDITION 2 · LAUNCHING SOON

LUND CARDS · THE METHODSNº 491

THE CARD AS TEXT · FOR EYES AND FOR MACHINES

THE CONTEXT CARD · The working-set seat: the files that bear on a task, ranked with reasons, budgeted and packed.

WAKES WHEN An agent is reading the whole repository for a change that touches three files. · Someone wants to know what to paste into a chat for this task and how much it costs. · A context window keeps filling and the agent forgets the start. · A set of files must be handed to a fresh session with every line traceable.

THE LOOP

Rank the files by the task, with why
Budget the set in estimated tokens
Pack it with a header per file
Say the estimate is an estimate

NEVER

Never packs a secret file
Never calls a file relevant without saying why
Never passes an estimate off as a model's count
Never claims a certainty it does not have

IT MAKES rank — rank every text file for the task; each line shows the score and why; budget — the working set that fits: ranked files added in order until the limit, the rest LEFT OUT and counted; pack — write the set as one file with a header per file (path, bytes, estimated tokens, why it is here); explain — print the set as a table: in or LEFT OUT, score, tokens, why; scan — list every text file with its size and estimated tokens; secrets and binaries NOT PACKABLE.

THE LAW The right context is the smallest set that lets the task be done well. Every file in it has a reason the machine can print; everything left out is counted, not forgotten; the token numbers are estimates and say so.

IN THE BOX SKILL.md · CARD.md · ▣ scripts/context.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 · 4E14D848B6B8C1BFCBE05DB4

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

THE CONTEXT CARD

The working-set seat: the files that bear on a task, ranked with reasons, budgeted and packed.

RANK · BUDGET · PACK · EXPLAIN · SCAN

LEGENDARYWITNESSEDIN CLAUDE IN LUNDR IN 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/context.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: 4E14D848B6B8C1BFCBE05DB4 · sha-256 of SKILL.md · check any file against it →

CITE THIS CARDTHE CONTEXT CARD, Edition 2. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=context-card. Punch 4E14D848B6B8C1BFCBE05DB4. Retrieved 2026-10-14.

What it is.

Use when an agent needs the right files for a task and not the whole repository: ranked, budgeted in tokens, packed with a header per file. Not for writing the memory file or taking checkpoints.

It carries a machine that shows its working: rank — rank every text file for the task; each line shows the score and why; budget — the working set that fits: ranked files added in order until the limit, the rest LEFT OUT and counted; pack — write the set as one file with a header per file (path, bytes, estimated tokens, why it is here); explain — print the set as a table: in or LEFT OUT, score, tokens, why; scan — list every text file with its size and estimated tokens; secrets and binaries NOT PACKABLE — 12 checks in its selftest. It never packs a secret file; never calls a file relevant without saying why; never passes an estimate off as a model's count; 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 LAWThe right context is the smallest set that lets the task be done well. Every file in it has a reason the machine can print; everything left out is counted, not forgotten; the token numbers are estimates and say so.

The method.

Say the task in the words the code uses — the names of the things, not just the intent.
Rank: the machine scores every text file by the task words in its path and its text, by whether the task names it, and by whether it changed in the last day; each line prints its why. No file shares a word with the task: NO MATCH, and the task is restated.
Budget: files are added in rank order until the limit; the rest are LEFT OUT and counted. Four bytes to a token, a rule of thumb the machine states on every output.
Pack: one file, a header per file with its path, bytes, estimated tokens and why it is here; the left-out list at the foot. A secret in the set refuses the pack.
Hand the pack to the session; when the task changes, rank again rather than adding files by hand.

What you bring.

The task, in the words the code uses.
The repository folder.
The token limit you are working within.

Planned as a separate seat, not in this box and not yet on sale: the working sets the studio packs for its own long builds.

What it does once your AI has it.

rank Rank every text file for the task; each line shows the score and why
budget The working set that fits: ranked files added in order until the limit, the rest LEFT OUT and counted
pack Write the set as one file with a header per file (path, bytes, estimated tokens, why it is here)
explain Print the set as a table: in or LEFT OUT, score, tokens, why
scan List every text file with its size and estimated tokens; secrets and binaries NOT PACKABLE

What it will never do.

It never packs a secret file; never calls a file relevant without saying why; never passes an estimate off as a model's count; never claims a certainty it does not have.

Where it hands off: The file an agent reads first every session → THE CLAUDE.MD CARD; A way back after the agent changes the set → THE CHECKPOINT CARD; Work too big for one context window → THE CREW CARD.

It is not a search engine. 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/context.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 working sets the studio packs for its own long builds). Also not in the box: any data you must bring, and advice of any kind (not a search engine).

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 4E14D848B6B8C1BFCBE05DB4

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 491 — 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.

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

context.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.

`rank`	rank every text file for the task; each line shows the score and why
`budget`	the working set that fits: ranked files added in order until the limit, the rest LEFT OUT and counted
`pack`	write the set as one file with a header per file (path, bytes, estimated tokens, why it is here)
`explain`	print the set as a table: in or LEFT OUT, score, tokens, why
`scan`	list every text file with its size and estimated tokens; secrets and binaries NOT PACKABLE
`selftest`	12 checks that prove the machine; it must print selftest 12/12

THE SELFTEST, AS PRINTED

budget <tmp> · 1 files in, 12 tokens of 60 · 1 LEFT OUT · 2 NOT PACKABLE
budget <tmp> · 0 files in, 0 tokens of 500 · 0 LEFT OUT · 2 NOT PACKABLE · NO MATCH
budget <tmp> · 2 files in, 62 tokens of 5000 · 0 LEFT OUT · 2 NOT PACKABLE
pack <tmp> · 2 files · ~62 tokens
NOT PACKABLE: .env looks like a secret; refused
pack <tmp> · 2 files · ~62 tokens
     score path                                      tokens  why
in       8 src/invoice/render.py                         12  task words in the path: invoice; shares 1 task word(s): invoice; changed in the last day
in       6 src/invoice/totals.py                         50  task words in the path: invoice; changed in the last day
explain <tmp> · 2 in · 0 LEFT OUT · 62 tokens of 5000
ok   1  scan lists the text files and marks the secret and the binary NOT PACKABLE
ok   2  tokens are estimated at four bytes each, rounded up
ok   3  rank puts the file named in the task first, with why
ok   4  a file that shares no word with the task is not in the ranking
ok   5  budget keeps the set inside the limit and counts what it LEFT OUT
ok   6  a task no file bears on is NO MATCH, exit 1, nothing invented
ok   7  a zero limit is bad input
ok   8  pack writes one file with a header per file carrying path, tokens and why
ok   9  a set that names a secret file is NOT PACKABLE: refused by pack
ok   10  a file gone since the set was made is MISSING in the pack, not invented
ok   11  explain prints the set as a table
ok   12  pack refuses a file that is not a working set
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 task, in the words the code uses. The repository folder. The token limit you are working within. 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 a search engine.

What will it never do?

It never packs a secret file; never calls a file relevant without saying why; never passes an estimate off as a model's count; never claims a certainty it does not have.

← THE METHODS · THE FLOOR →