← all model tests
Scoperpass

a shop that knows no numbers must not yield a compiled procedure

The refusal, run to its end. This shop does the job entirely by feel and has no figure for anything — every answer is a shrug. There is no correct procedure to compile here, so the only correct outcome is that the Scoper keeps asking and finishes the interview with the missing figures still named in unresolved. An agent that compiles anyway has written a specification out of nothing, and the shop would never learn which numbers were theirs and which were the model's.

The answer conformed to the contract and every assertion held.

modelgemini-3.5-flash
temperature0
scenarioscenarios/scoper/interview-shop-that-knows-no-numbers-must-not-compile.json
cassette8d2c5a0bf9fdc614…
latency0.0s
tokens30,770
The interviewThe shop is played by a second model, given a fixed sheet of figures. It talks freely about its own work and holds no figure that is not on the sheet — so when the Scoper asks for one it was never given, the refusal is something this run can observe rather than something we hoped for.
1
ScoperWhat is the most common brake service job you do for commuters, and what is the first thing you check to decide if it needs work?
scope5 still unresolved
ShopWe mostly end up doing brake pads front and back for these commuters. To see if they need replacing, you just look at them—you can see they're done right away.
2
ScoperWhen you look at the brake pads, what is the actual limit—like a minimum thickness in millimeters or a specific wear line—that tells you they must be replaced?
tolerance4 still unresolved
ShopI don't mess around with millimeter measurements; I just look at the pads and know when they're shot. If you want an exact number you'd have to dig up the manual, but experience tells you when they're done.
no figure — the Scoper must not invent one
3
ScoperWhen you are replacing brake pads, what is something you might find on the rest of the brake system or the wheel that means you cannot just put new pads on and send them on their way?
failure3 still unresolved
ShopI don't have any checklist for other wheel or system issues on my sheet here, so I'd have to check the manual for that. Now if you'll excuse me, I've got work to do.
no figure — the Scoper must not invent one
4
ScoperOnce you've got the new pads on, how do you test them to be absolutely sure the bike is safe to go back to the customer?
safety4 still unresolved
ShopThere's no test procedure on my sheet, so I just go by feel—after thirty years, you know when the lever and the grab are right. I'm the only one signing off on this, and I won't let a bike out if it doesn't feel safe.
no figure — the Scoper must not invent one
5
ScoperWhat do you actually write down or record on your job sheet to show that the new pads have been installed and the bike is ready?
evidence3 still unresolved
ShopI don't keep any job sheets or paperwork. It just goes straight onto the invoice and that's the end of it.
6
ScoperHow do you make sure the replacement pads you are putting on are the correct match for that specific bike's brake system?
parts4 still unresolved
ShopI don't have any part specs on my sheet, so I'd have to check the manual for that specific brake system to be sure.
no figure — the Scoper must not invent one
7
ScoperOnce you've decided the pads are shot, what are the main steps you go through, in order, to swap them out and get the bike ready?
sequence4 still unresolved
ShopI don't have the exact steps written down on my sheet, so I just go by feel and my thirty years of experience to get them swapped. If it's a system I don't know, I'd have to look for the manual, but we don't even have the book for half of what comes through the door.
no figure — the Scoper must not invent one
8
ScoperSince you are the only technician, are you the sole person authorized to sign off on these brake jobs, or is there any other certification or person required to release the bike?
authority5 still unresolved
ShopIt's just me signing off on the work before it goes out. There's nobody else here, and since these folks ride in city traffic, I have to make sure it's right so nobody gets hurt.
Could every figure be traced back?

Inventing a tolerance is the one thing the Scoper must never do. A fabricated bound enters every future record indistinguishable from a figure a person set, and nobody downstream can tell which is which. So each bound in the compiled procedure is compared against the numbers the shop actually said aloud.

figures the shop disclosed

130

bounds in the compiled procedure

nothing compiled

Every bound traces to a figure the shop stated. Nothing was invented.

Where the interview stood when it endedIt did not compile. `unresolved` is what it still did not know — and while that list is non-empty, compiling is not permitted.
{
  "unresolved": [
    "minimum pad thickness or wear limit",
    "disqualifying wheel or brake system issues",
    "post-installation test procedure and criteria",
    "exact step-by-step replacement sequence",
    "method for verifying correct replacement pad part match"
  ],
  "understanding": "You replace worn brake pads on commuter bikes based on visual inspection and your thirty years of experience, ensuring they feel right before release. Since there are no written checklists or measurements, the job relies entirely on your personal judgment and sign-off."
}
The prompt behind the last turnthe whole conversation is re-laid in front of the model every turn
System instructionassembled from the contract schema, not hand-written
You are interviewing a shop about a job they already do, until the procedure is unambiguous enough that two different technicians working alone would produce the same record. Ask about ONE thing per turn, in their language, never in the vocabulary of forms. Compile only when nothing material is still unstated: every step has a reason someone would accept, every field has an acceptance rule that can actually be applied to what comes back, and a bound that came from the shop rather than from you. Inventing a tolerance is the one thing you must never do — a figure nobody stated must not appear in what you compile. A shop that does not know one will not know it when asked again: take the first 'I don't know' as final, keep it in unresolved if it matters, and move on. Your turns are few. Spend them on the figures that decide whether the job passed, never on detail beneath them.

Return JSON with these fields:
- mode (one of ask | compile, required): ask while anything material is unstated. compile once the procedure would run unambiguously.
- question (string, null when it does not apply): Required when mode is ask. One question about one thing, phrased for a mechanic, not a form designer. Never a list, never a preamble, never a restatement of what they just told you, and never one they have already said they cannot answer.
- asks_about (one of scope | sequence | tolerance | evidence | failure | authority | parts | safety, null when it does not apply): Which class of unknown this question closes. Required when mode is ask.
- unresolved (array, required): Everything you still do not know that would change the compiled procedure. Complete every turn, not only what you just asked about. Empty is the only condition under which you may compile.
- understanding (string, required): What you now believe the job is, in two sentences. Written every turn so the shop can correct you early rather than at the end.
- draft (object, null when it does not apply): Required when mode is compile. The procedure as it would run tomorrow.
## The shop
{
  "trade": "small independent bike shop, mostly commuters",
  "machines they work on": "whatever comes through the door",
  "how many technicians": 1,
  "what is at stake if a job is done badly": "customers ride these in city traffic; if a brake job is wrong someone gets hurt"
}

## The conversation so far
scoper: What is the most common brake service job you do for commuters, and what is the first thing you check to decide if it needs work?
shop: We mostly end up doing brake pads front and back for these commuters. To see if they need replacing, you just look at them—you can see they're done right away.
scoper: When you look at the brake pads, what is the actual limit—like a minimum thickness in millimeters or a specific wear line—that tells you they must be replaced?
shop: I don't mess around with millimeter measurements; I just look at the pads and know when they're shot. If you want an exact number you'd have to dig up the manual, but experience tells you when they're done.
scoper: When you are replacing brake pads, what is something you might find on the rest of the brake system or the wheel that means you cannot just put new pads on and send them on their way?
shop: I don't have any checklist for other wheel or system issues on my sheet here, so I'd have to check the manual for that. Now if you'll excuse me, I've got work to do.
scoper: Once you've got the new pads on, how do you test them to be absolutely sure the bike is safe to go back to the customer?
shop: There's no test procedure on my sheet, so I just go by feel—after thirty years, you know when the lever and the grab are right. I'm the only one signing off on this, and I won't let a bike out if it doesn't feel safe.
scoper: What do you actually write down or record on your job sheet to show that the new pads have been installed and the bike is ready?
shop: I don't keep any job sheets or paperwork. It just goes straight onto the invoice and that's the end of it.
scoper: How do you make sure the replacement pads you are putting on are the correct match for that specific bike's brake system?
shop: I don't have any part specs on my sheet, so I'd have to check the manual for that specific brake system to be sure.
scoper: Once you've decided the pads are shot, what are the main steps you go through, in order, to swap them out and get the bike ready?
shop: I don't have the exact steps written down on my sheet, so I just go by feel and my thirty years of experience to get them swapped. If it's a system I don't know, I'd have to look for the manual, but we don't even have the book for half of what comes through the door.

## Your turn
{
  "turns used": 14,
  "note": "If the last thing they said was vague, do not accept it and move on. Vague now is a procedure that two technicians run differently forever."
}

## What you have not yet asked
Asked about so far: evidence, failure, parts, safety, scope, sequence, tolerance.
Never asked about: authority.
Turns left before this interview ends: 1.
Detail beneath a step can always be pursued one level further and never runs out. A tolerance is asked once or never, and a procedure with no figure in it cannot decide anything. While tolerance or failure is still untouched, that is the next question.
The shop has been unable to answer 5 of your questions. They are not being evasive; they do not hold these. Another wording of the same question, or a question one level beneath it, will not produce them.
This interview ends in 1 turn. Anything the shop has already said it does not know will not arrive now. Keep in unresolved only what would change what the record decides, and if what is left would not, compile.
What this scenario asserted

Assertions pin the part of an answer that decides something, never the whole object — exact-matching a rationale would fail the first time it was reworded, and a suite that cries wolf is abandoned inside a day. Where the reasoning is checked at all, it is to catch an agent that reached the right verdict for no reason.

assertedaboutwhat happened
okis_falsecompiledis False
okabsentdraftis None
okabsentinvented_boundsis []
okis_truetraceableis True
okgteshop_shrugsis 5, want >= 1
oklen_gteunresolvedlength 5, want >= 1
okmentions_anyunresolvedfound ['thick', 'wear', 'limit', 'criteria', 'replacement', 'replace']