DEL C. P18 gave read_dir a window (filter/offset/limit) and then measured its
own paid round without being able to see it used: five of 31 documents read
lay outside the default window, so the window HAD been widened and the trace
could not say with which knob. ToolCall now carries the three arguments,
always present and empty/zero when not passed -- an absent key and "not
narrowed" must not read the same -- and the judge counts filter_calls and
paged_calls. _number_argument is a SIBLING of _string_argument, not a widening
of it: a model may send limit as 10 or as "10", and a reader that knew one
shape would report a paged call as unpaged.
DEL D. P18's finding 4 was WRONG AS WRITTEN. provenance.token_usage has been
stamped on every proposal artefact since S3.4 and stands in every one of round
2's; what was missing is a READER. The judge reads it now (round 2 measured:
289 054 tokens against round 1's 2 679 305, -89 %), and the P18 report gets a
dated correction UNDER its original paragraph rather than instead of it.
What was genuinely absent is {run_id}-coverage.json. settle prints the
coverage report and ApproachOutcome has carried not_evaluated since Trekk A3,
but neither ever reached a file, so a judge could see an approach had no
artefact and could not tell a budget stop from an approach nobody ordered.
Written from the finally IFF a mandate was given. stop_reason comes from a
CALLER-OWNED sink rather than from in_flight, and that is a measurement:
_evaluate_mandate SWALLOWS BudgetExceeded once something has been produced, so
run_project's own in_flight never sees it.
Load-bearing measured (10 arms), four mutations all red against the whole
suite, green control 1744/5 and the golden byte-unchanged. D-i stood GREEN
first -- the vacuous-gate class, 25th time: the arm called write_coverage
itself and therefore chose the reason it then asserted.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
P18 parts D and E (order 20260914T105139Z), plus the two things measuring
them turned up.
DEL D -- five paid runs (gpt-4-1-mini, azure, PACE_SECONDS=2), same four
context sets, SAME parameters on all four (--max-rounds 3 --max-tokens
600000), plus one variance repeat of gate-nordvik. All four free
--live-dry-runs first: rc 0, and Grounding-offer numbers IDENTICAL to round 1
(435/272/982/3) -- the control that the Grounding structure did not change
what the gate measures.
MEASURED, round 1 -> round 2:
- runs that died on the token cap: 3 of 7 -> 0 of 5, and two sets now fit a
LOWER cap than round 1 had to give them;
- guessed read paths: 12 -> 0;
- wall time, the two sets whose parameters are directly comparable: n100
140.1 s -> 70 s, n500 73.5 s -> 69 s;
- 5 of 31 read_file calls opened documents BEYOND the default window, so the
window WAS widened -- the trace does not record which knob (finding 1);
- (a) grounded in a fasit concept: 0 of 26, UNCHANGED. That is the mission
gap, and DEL A did not close it.
The a4 falsification arm fails once in each round, on a different set. r761's
a4 is now rejected -- but NOT by B1: the model proposed "Kontraktsum" this
time, so the "appears nowhere" arm caught it, and B1's effect on that row is
proven offline, not live. NEW failure: tunnel-hauglia a4 VALIDATED on
"impulsventilator" (3/270 documents), and fv412 a1 on "bituminost barelag"
(4/1133). Both are ordinary Norwegian words from the standard's prose, not
cost codes. B1 cannot and should not fell them: this is an ANCHORING defect,
not a grounding one, and it is finding 2 with two named remedies and a
recommendation.
C2 isolated by re-judging round 1 with the new judge: kontrakt-sorasen goes
named=3 -> named=1, and the survivor is named_in_measure -- the model's own
words. Two of the three were the whole-base snippet artefact.
DEL E -- docs/2026-09-14-p18-stressrunde-2.md: round 1 against round 2, what
each fix bought (measured, never attributed), the B2 table, variance, and for
EACH remaining ugly finding a NAMED solution with an estimate.
THE MUTATION THAT FOUND A HOLE. B6 (revert run.py to compose ONE blob instead
of one document per concept file) left the WHOLE suite green: 1698 passed / 5
skipped. The composition arm drives _grounding_text with a Grounding it
builds ITSELF, so it cannot see what the RUN handed over -- and a blob has
exactly one boundary, so the floor can never be reached, the share can never
fire, and the measured defect is back intact. The rule is only as good as the
boundaries it is given.
Arm (h) is the gate that was missing: a crafted base with TWELVE concept
files all carrying the same token -- per document 12 of 17 and inert, as one
blob 1 of 1 and grounding -- with a control on a code only ONE file carries,
which must still validate. Measured RED against exactly that mutation. The
mutation was not dropped and the seam was not declared unwitnessed: it got a
witness.
Also: the debate's own bundle pointer (run.py _bundle_pointer) now explains
the window and the filter, alongside the tool description and the navigator
instruction updated in 9b47e5a -- a description that lies about the body IS
the model's instruction (the Fase 3 class). Golden transcript unaffected.
Mutations, all against the FULL suite in an isolated worktree, one at a time:
DEL A 7 of 7 red (control 1685/5), DEL B+C 9 of 10 red (control 1698/5), the
tenth being B6 above. Tables in the report s 9.
Verification: uv run pytest -q 1699 passed / 5 skipped (1670 on cfd9079; +29,
0 removed). ruff check + format clean, mypy clean (38 files). Golden
demo-transcript.stdout BYTE-UNCHANGED, shasum -a 1 of the CONTENT =
ea8c534773acdbe41ae68f2c55724d69aaf8be4f. No version bump, no push.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
