THE BENCHMARK CARD · the client beside its peers, sourced
WAKES WHEN a client asks how they compare, or a benchmark is built for a recommendation
THE LOOP
gather: every figure with source and year
sources · the unsourced and the stale
table · quartiles, rank, direction
gap · to the median, to the top quartile
say where the ground is to make up
NEVER
Guess which way a metric is better
Read a blank as zero
Use a figure with no source
IT MAKES table — per metric: the client, rank, Q1, median, Q3; gap — the gap to the median and the top quartile, units and percent; sources — every row's source and year; the unsourced and the stale.
THE STANDARD every figure is sourced and dated; direction is written, never guessed; a blank is a blank; the top quartile depends on direction; gaps are signed one way; only what may be used.
PROOF SELFTEST 8/8 · 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 · 7897ACAB516AC4C676F07068
CITE THIS CARDTHE BENCHMARK CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=benchmark-card. Punch 7897ACAB516AC4C676F07068. Retrieved 2026-10-10.
What it is.
Sets a client beside its peers with every figure on the record: table reads a table of peers by metrics — each column marked (+) when higher is better or (-) when lower is — and gives each metric's quartiles, median and the client's rank, leaving blank cells out rather than reading them as zero; gap gives the distance from the client to the median and to the top quartile, in units and percent, signed so positive is ground to make up; sources reads every row for its source and year, naming the unsourced and the older than a given year.
It carries a machine that shows its working: table — per metric: the client, rank, Q1, median, Q3; gap — the gap to the median and the top quartile, units and percent; sources — every row's source and year; the unsourced and the stale — 8 checks in its selftest, every command on the record. It will never guess which way a metric is better; never read a blank as zero; never use a figure with no source.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
EVERY FIGURE IS SOURCED AND DATEDA row with no source or no year is named, and a stale year too.
DIRECTION IS WRITTEN, NEVER GUESSEDEach metric is marked (+) or (-); a column without the mark is refused.
A BLANK IS A BLANKA missing figure is left out of the quartiles, not counted as zero.
THE TOP QUARTILE DEPENDS ON DIRECTIONFor lower-is-better metrics the top is the first quartile.
GAPS ARE SIGNED ONE WAYPositive is ground to make up; zero or negative is at or past the line.
ONLY WHAT MAY BE USEDPeers' figures come from what they publish or what the client is entitled to share; the card fetches nothing.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/THE-CONSULTANCY.md
THE MACHINEbenchmark.py · three commandstable — per metric: the client, rank, Q1, median, Q3; gap — the gap to the median and the top quartile, units and percent; sources — every row's source and year; the unsourced and the stale. Every command prints its working and refuses a bad input; 8 checks in its selftest, every command on the record.
What it does once your AI has it.
table Per metric: the client, rank, Q1, median, Q3
gap The gap to the median and the top quartile, units and percent
sources Every row's source and year; the unsourced and the stale
On the record The sources read first — one peer named for no source and a stale year; the table gave each metric its quartiles and the client's rank, the blank left out; the gap put the client four points of waste from the median and five from the top quartile.
What it will never do.
It will never guess which way a metric is better; never read a blank as zero; never use a figure with no source.
Quartiles of a handful of peers are rough; with fewer than four figures the card reports the lowest, the median and the highest. Peers measured differently — a different year, a different definition — are not comparable, and the card cannot see that; the source column is there so a reader can. Not investment advice. 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-CONSULTANCY.md — the guide in references/, in the card's own voice
scripts/benchmark.py — the machine: table · gap · sources, standard library only, selftest 8/8
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 8/8 · 3 COMMANDS ON THE RECORD · WITNESSED IN CLAUDE 2026-10-10 · EDITION 1 · PUNCH 7897ACAB516AC4C676F07068
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 1104 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
benchmark.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.
table
per metric: the client, rank, Q1, median, Q3
gap
the gap to the median and the top quartile, units and percent
sources
every row's source and year; the unsourced and the stale
selftest
8 checks that prove the machine; it must print 8/8
(the client and four peers, each figure with its source and year — one row without)
$ (the client and four peers, each figure with its source and year — one row without)
sources · EVERY ROW'S SOURCE AND YEAR; THE UNSOURCED AND THE STALE
$ python3 scripts/benchmark.py sources peers.csv --since 2023
Northfield own books 2026
Peer A annual report 2025
Peer B trade survey 2025
Peer C — 2021
Peer D trade survey 2025
NO SOURCE · Peer C
STALE (2021, before 2023) · Peer C
5 rows · 1 without a source or a good year
table · PER METRIC: THE CLIENT, RANK, Q1, MEDIAN, Q3
$ python3 scripts/benchmark.py table peers.csv --us Northfield
waste % ↓ us 14 rank 5 of 5 · Q1 9 · median 10 · Q3 12
sales per hour ↑ us 310 rank 3 of 4 · Q1 305 · median 345 · Q3 390
2 metrics · 4 peers and Northfield · ↑ higher is better, ↓ lower
gap · THE GAP TO THE MEDIAN AND THE TOP QUARTILE, UNITS AND PERCENT
$ python3 scripts/benchmark.py gap peers.csv --us Northfield
waste % to the median +4 (+29%) · to the top quartile +5 (+36%)
sales per hour to the median +35 (+11%) · to the top quartile +80 (+26%)
positive is ground to make up; zero or negative is at or past the line
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?
Gather: every figure with source and year. It asks first for the peers, the metrics with their direction, and each figure's source and year.
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 (8/8) 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. Quartiles of a handful of peers are rough; with fewer than four figures the card reports the lowest, the median and the highest. Peers measured differently — a different year, a different definition — are not comparable, and the card cannot see that; the source column is there so a reader can. Not investment advice.
What will it never do?
It will never guess which way a metric is better; never read a blank as zero; never use a figure with no source.
