THE FRESH EYES · a second agent reviews the work, not the way
WAKES WHEN work from one agent needs a second reader who has not seen how it was made
THE LOOP
fresheyes.py pack · brief, work, rubric
the reviewer reads and runs; sees no notes
a mark and a quote for each of the five
verdict · PASS, FIX, STOP, or incomplete
FIX → the changes go back; review again
NEVER
Show the reviewer the first agent's reasoning
Count a mark with no quote
Average away a zero on SAFE
IT MAKES pack — the reviewer's folder, the reasoning left out and listed; rubric — the five criteria, a mark and a quote each; verdict — the filled rubric read: the verdict and the changes.
THE STANDARD the reviewer sees the work, not the way; five criteria, every time; a mark needs a quote; sAFE cannot be averaged; one of four verdicts; fIX goes back and comes back.
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 · D8BECD228E0A8261F8DC7145
CITE THIS CARDTHE FRESH EYES, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=fresh-eyes-card. Punch D8BECD228E0A8261F8DC7145. Retrieved 2026-10-10.
What it is.
Runs the fresh-eyes review: packs a reviewer's folder with the brief, the work and a fixed five-part rubric while leaving out the first agent's notes, scratch, drafts and reasoning (listed, so nothing is hidden), and reads the filled rubric into a verdict — PASS, FIX with the changes named, STOP, or INCOMPLETE REVIEW when a mark has no quote behind it.
It carries a machine that shows its working: pack — the reviewer's folder, the reasoning left out and listed; rubric — the five criteria, a mark and a quote each; verdict — the filled rubric read: the verdict and the changes — 10 checks in its selftest, every command on the record. It will never show the reviewer the first agent's reasoning; never count a mark with no quote; never average away a zero on SAFE.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
THE REVIEWER SEES THE WORK, NOT THE WAYNotes, scratch and reasoning stay out, and the list of what stayed out goes in.
FIVE CRITERIA, EVERY TIMECorrect, complete, evidenced, safe, clear.
A MARK NEEDS A QUOTEThe line, the file or the run the reviewer saw; without it the mark does not count.
SAFE CANNOT BE AVERAGEDA zero on SAFE is STOP whatever the other four say.
ONE OF FOUR VERDICTSOne of four verdicts, in one shape, so the sender never has to interpret a review.
FIX GOES BACK AND COMES BACKA review is a loop until PASS, not an opinion delivered once.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/WHY-THE-REVIEWER-SEES-LESS.md
THE MACHINEfresheyes.py · three commandspack — the reviewer's folder, the reasoning left out and listed; rubric — the five criteria, a mark and a quote each; verdict — the filled rubric read: the verdict and the changes. 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.
pack The reviewer's folder, the reasoning left out and listed
rubric The five criteria, a mark and a quote each
verdict The filled rubric read: the verdict and the changes
On the record A work folder packed for review with the brief and the rubric, the notes folder and the scratch file left out and listed; the rubric written; the reviewer's filled rubric read into FIX, 9 of 10, with the one change named: say which file is the deliverable.
What it will never do.
It will never show the reviewer the first agent's reasoning; never count a mark with no quote; never average away a zero on SAFE.
Reasoning is left out by file-name pattern (notes, scratch, thinking, drafts, logs…); reasoning kept under another name travels with the work. The verdict reads marks and quotes; it cannot judge whether a quote is apt. The reviewer's independence is the sender's to arrange. 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-THE-REVIEWER-SEES-LESS.md — the guide in references/, in the card's own voice
scripts/fresheyes.py — the machine: pack · rubric · verdict, 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 D8BECD228E0A8261F8DC7145
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 911 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
fresheyes.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.
pack
the reviewer's folder, the reasoning left out and listed
rubric
the five criteria, a mark and a quote each
verdict
the filled rubric read: the verdict and the changes
selftest
10 checks that prove the machine; it must print 10/10
pack · THE REVIEWER'S FOLDER, THE REASONING LEFT OUT AND LISTED
$ python3 scripts/fresheyes.py pack work --brief brief.md --out review
review folder review · 2 files in · 2 left out (LEFT-OUT.md)
cat review/LEFT-OUT.md
$ cat review/LEFT-OUT.md
# Left out of the review folder, on purpose
The first agent's reasoning is not here, so the review cannot be anchored on it:
- notes/
- scratch.txt
rubric · THE FIVE CRITERIA, A MARK AND A QUOTE EACH
$ python3 scripts/fresheyes.py rubric --out review/rubric-blank.md
wrote review/rubric-blank.md
(the reviewer, who never saw notes/thinking.md, fills the rubric: four twos and a one on CLEAR, each with its
$ (the reviewer, who never saw notes/thinking.md, fills the rubric: four twos and a one on CLEAR, each with its quote)
verdict · THE FILLED RUBRIC READ: THE VERDICT AND THE CHANGES
$ python3 scripts/fresheyes.py verdict review/rubric.md
FIX · 9/10 · CORRECT 2 · COMPLETE 2 · EVIDENCED 2 · SAFE 2 · CLEAR 1
must change: say in REPORT.md which file is the deliverable
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?
Fresheyes.py pack · brief, work, rubric. It asks first for the brief the work was done to, the work folder, and a reviewer who has not been in the room.
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. Reasoning is left out by file-name pattern (notes, scratch, thinking, drafts, logs…); reasoning kept under another name travels with the work. The verdict reads marks and quotes; it cannot judge whether a quote is apt. The reviewer's independence is the sender's to arrange.
What will it never do?
It will never show the reviewer the first agent's reasoning; never count a mark with no quote; never average away a zero on SAFE.
