THE READING SHELF · books served to an AI by retrieval
WAKES WHEN someone wants an AI to answer from their own books and say where the answer came from
THE LOOP
index <folder> · the shelf and its map
ask "<question>" · passages, cited
read the passages; quote with the citation
cite <book> <n> · one passage whole
nothing found: say so; never from memory
NEVER
Answer from memory when the shelf is silent
Quote without the book and the passage
Send a book anywhere
IT MAKES index — every .md and .txt in the folder is a book; the passages and the map written; ask — the passages that carry the question's words, scored, each cited book · passage; cite — one passage whole, with its citation, to quote from; map — the shelf: every book, its passages, its size.
THE STANDARD retrieval, not training; every answer is cited; silence is an answer; passages stop at paragraph edges; the map is public to its owner; the books stay where they are.
PROOF SELFTEST 12/12 · witnessed in Claude 2026-10-10 · 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 · B42B4CAC0893E68A1D471E4C
CITE THIS CARDTHE READING SHELF, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=reading-shelf-card. Punch B42B4CAC0893E68A1D471E4C. Retrieved 2026-10-10.
What it is.
Serves a shelf of books to an AI by retrieval instead of training: index cuts every .md and .txt in a folder into passages at paragraph edges and writes the shelf and its map; ask scores the passages that carry a question's words and returns the best few, each cited book and passage, with the words that matched; cite returns one passage whole to quote from; map lists every book with its passages and size. Nothing is learned from the books and nothing leaves the computer.
It carries a machine that shows its working: index — every .md and .txt in the folder is a book; the passages and the map written; ask — the passages that carry the question's words, scored, each cited book · passage; cite — one passage whole, with its citation, to quote from; map — the shelf: every book, its passages, its size — 12 checks in its selftest, every command on the record. It will never answer from memory when the shelf is silent; never quote without the book and the passage; never send a book anywhere.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
RETRIEVAL, NOT TRAININGThe books are cut and indexed; nothing is learned from them, and the model is the same after as before.
EVERY ANSWER IS CITEDBook and passage — or it is not an answer from the shelf.
SILENCE IS AN ANSWERWhen no passage carries the question's words the card says so, and the AI must not fill the gap from memory.
PASSAGES STOP AT PARAGRAPH EDGESPassages stop at paragraph edges, so a quote is never cut mid-sentence; a long paragraph stands whole.
THE MAP IS PUBLIC TO ITS OWNEREvery book, its file, its passages, its size, written beside the shelf.
THE BOOKS STAY WHERE THEY AREThe shelf is a file beside them; nothing is copied anywhere else and nothing is sent.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/WHY-RETRIEVAL.md
THE MACHINEshelve.py · four commandsindex — every .md and .txt in the folder is a book; the passages and the map written; ask — the passages that carry the question's words, scored, each cited book · passage; cite — one passage whole, with its citation, to quote from; map — the shelf: every book, its passages, its size. 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.
index Every .md and .txt in the folder is a book; the passages and the map written
ask The passages that carry the question's words, scored, each cited book · passage
cite One passage whole, with its citation, to quote from
map The shelf: every book, its passages, its size
On the record Two books shelved and mapped; a question found the tide passage and cited it; one passage returned whole to quote from; a question with no words on the shelf was answered with silence, named as such.
What it will never do.
It will never answer from memory when the shelf is silent; never quote without the book and the passage; never send a book anywhere.
The shelf matches words, weighted by how rare they are on the shelf; a question in other words than the book's can miss, and a passage that carries the words can still not answer. The AI reads what comes back and says what it says; the card does not understand the books. Plain text only: a PDF or a Word file is a book once its words are in a .md or .txt. 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
WHY-RETRIEVAL.md — the guide in references/, in the card's own voice
scripts/shelve.py — the machine: index · ask · cite · map, 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 2026-10-10 · EDITION 1 · PUNCH B42B4CAC0893E68A1D471E4C
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 1086 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 2026-10-10
shelve.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.
index
every .md and .txt in the folder is a book; the passages and the map written
ask
the passages that carry the question's words, scored, each cited book · passage
cite
one passage whole, with its citation, to quote from
map
the shelf: every book, its passages, its size
selftest
12 checks that prove the machine; it must print 12/12
(two small books written into books/)
$ (two small books written into books/)
index · EVERY .MD AND .TXT IN THE FOLDER IS A BOOK; THE PASSAGES AND THE MAP WRITTEN
$ python3 scripts/shelve.py index books --size 120
shelved 2 books · 4 passages · books/shelf.json · the map at books/SHELF-MAP.md
map · THE SHELF: EVERY BOOK, ITS PASSAGES, ITS SIZE
ask · THE PASSAGES THAT CARRY THE QUESTION'S WORDS, SCORED, EACH CITED BOOK · PASSAGE
$ python3 scripts/shelve.py ask when does the spring tide come? --shelf books/shelf.json --k 2
— The Book of Tides · passage 1 · score 1.392 · words: spring, tide
# The Book of Tides The spring tide comes with the new moon and the full moon, when the sun and the moon pull together.
— The Book of Tides · passage 2 · score 0.613 · words: tide
The neap tide comes at the quarters, when they pull across each other and the water barely moves.
2 passages · quote with the citation (book · passage), never from memory
cite · ONE PASSAGE WHOLE, WITH ITS CITATION, TO QUOTE FROM
$ python3 scripts/shelve.py cite --shelf books/shelf.json The Book of Tides 2
The Book of Tides · passage 2
The neap tide comes at the quarters, when they pull across each other and the water barely moves.
ask · THE PASSAGES THAT CARRY THE QUESTION'S WORDS, SCORED, EACH CITED BOOK · PASSAGE
$ python3 scripts/shelve.py ask quantum chromodynamics --shelf books/shelf.json
nothing on the shelf carries those words; the AI should say so rather than answer from memory
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?
Index <folder> · the shelf and its map. It asks first for the folder of books, and the question.
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. The shelf matches words, weighted by how rare they are on the shelf; a question in other words than the book's can miss, and a passage that carries the words can still not answer. The AI reads what comes back and says what it says; the card does not understand the books. Plain text only: a PDF or a Word file is a book once its words are in a .md or .txt.
What will it never do?
It will never answer from memory when the shelf is silent; never quote without the book and the passage; never send a book anywhere.
