Four of the maintainer's five playtest remarks. Three of the five turned
out to be data the projection already carried, rendered as text -- the
table's problem was legibility, not content, and the coverage gate passes
either way because it proves nothing is OMITTED, not that anything is
readable. That gap is named in the evidence rather than closed: the honest
control is a person playing it.
T01. The ending control was two defects wearing one button. The label said
"close -- I have read this" while hotseat.rs reads `done` as STOP THE
SERVER, and acknowledging it changed nothing -- the tab kept a full table
and a `play again` pointing at a closed port. Now labelled by its effect,
and the page seals itself on the `closed` reply: removeAttribute on every
control's data-drop, so they stop being droppable by the same rule that
made them droppable. removeAttribute rather than setAttribute(_, null),
which writes the truthy string "null" in a browser.
The reason it survived is structural. jsrun's fetch stub returned
{then: function(){return this}} and never invoked its callbacks, so every
line of the script reacting to the server was unreachable from every test
in this project -- a page that ignores the server was indistinguishable
from one that acts on it. Same finding as CB-WP-0016's "a stub too thin to
express a failure is how the failure survives", one layer deeper, at the
reply. The stub now delivers a real then-chain; gesture_with_reply reports
surviving controls; the seal is mutation-proven and a negative control
asserts `ok: dealing` does NOT seal.
T02. Draw and discard as offset stacks with counts. The shuffle question
the task required settling: it already works, at
games/ground/src/lib.rs:1419-1435, implementing the U4 default that
ground-game confirmed 2026-08-03. Nothing raised. The piles show the state
before it fires, which is derivable from the view; a claim that a
reshuffle HAS happened is not, and is not made. CB-WP-0026 applied that
ruling the same day this consumed it -- first time answering "is this
underdetermined?" was one lookup instead of a message.
T03. Each seat's play drawn as a card, sentence kept beside it. The
face-down back is a const with no parameters: SelectionView::Hidden
carries nothing, so there is no data path into the back to add later. The
leak test copies view.rs's own shape -- identical backs across two
different hidden situations, THEN assert a revealed play does show,
because without the second half the first passes for a renderer that draws
nothing.
T04. MatchTally lives in `play`, beside the listener and the seed. What
"cumulative" means was decided before anything was summed, and the answer
is that GROUND defines one game and no series: summed personal score and
games-won answer different questions, and a test asserts they can point at
different seats. Both shown, both labelled. Registered F15 as a NOTE --
the test shows the tallies can differ, which is arithmetic, not evidence
the ambiguity harms play, so GameDesign §3.1 bars reporting it. First use
of the note tier since D6 wrote it, and it came from building rather than
from play.
make all: exit 0. 41 render tests, 26 cb-play tests, loop-lint clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
ground-game ruled all ten U-items on 2026-08-03, every one CONFIRMED as
the default clay-borg simulates, and confirmed five of six provisional
scenarios. clay-borg never collected the answers: CB-RES-0007 reported "0
of 10 ruled" the same day, and CB-WP-0022 built the finding register two
days later still recording them as `reported`. make design's first run is
what noticed -- not a human, not the adversarial review that found four
other things.
That is the unread-inbox failure running in the opposite direction, and it
appears nowhere in the declaration, survey, ADR or spec of the pass that
was built entirely around the forward version. It is arguably worse: an
unread message is visible as silence, while a collected-but-unapplied
ruling looks exactly like work in progress.
Ten rulings quoted into §Underdetermined (the three conditional ones
verbatim -- U1's designer note, U2's End-only trigger, U8's
consume-only-if-it-cancels). Five provisional flags lifted, replaced by
ruled/ruled_by/ruled_note so the flag went and the provenance stayed.
Register queue 9 -> 0.
T02's control came back clean: make sim is 26 passed, 59 rules covered,
nothing red. Had a scenario gone red it would have meant we described our
own behaviour incorrectly to ground-game.
I wrote two U-item mappings and both were wrong. gr-a04 -> U1 (it asserts
consent is REQUIRED; U1 asks WHEN the target accepts) and gr-d05 -> U5 (it
exercises the UNREJECTED Reverse; U5 is the rejected one). Both plausible
from covers:, neither survived reading the description. Third and fourth
instance of this defect; the first two reached ground-game. So encodes_u_item
is now a declaration and design.py asserts the file names what it claims --
and that check's own first version grepped for mentions and went red when
two files recorded why they do NOT encode U1 and U5. A mention is not a
claim, which is exactly the looseness that let "six of the ten have
provisional scenarios" stand.
Two positive controls went red for the best possible reason, both broken
the same way -- asserting against live repo data instead of constructing
their condition. rule-coverage.py required at least one provisional item
to EXIST; it now builds a fixture and reports the live count as a
diagnostic, because there is no number of provisional items this project
should have. design-baseline.py pinned "2 of 6" while recomputing one row
from a live glob, so the dated snapshot was never a snapshot; frozen to
its 2026-08-03 list and unwired from self-tests, since per ADR-0012 D8 it
is no longer a reporting tool.
ScenarioFile is deny_unknown_fields and refused the four new fields until
declared -- correct: a corpus accepting unknown metadata would let a typo'd
encodes_u_iem sit there claiming nothing.
DEVIATION: ADR-0012 D2 said "no new file". GroundRules.md crossed the
loadability limit, so the register moved to specs/FindingRegister.md. D2's
substance holds -- one register, same machinery, nothing competing -- but
the literal instruction did not, and it resolves an awkwardness D2 named
itself.
make all: exit 0. loop-lint clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
T05. tools/design.py, make design, and the register in GroundRules.md --
14 rows, no new file, because ADR-0012 D2 made §Underdetermined the
register rather than building one beside it.
Backfill was the test and it caught two things the ADR did not have.
First, a `role` column. The first report alarmed on U2 and was wrong to:
U2's scenario is green BECAUSE the provisional default it documents is
implemented, which says nothing about whether ground-game agrees. GR-E01's
was a counterexample that went green. Same colour, opposite meaning -- a
register that cannot tell them apart either alarms constantly or never.
Only a green counterexample alarms. Folded back into GameDesign §1.3.
Second, it contradicted the survey. CB-RES-0007 said "six of the ten
already have provisional scenarios." Measured -- grep -lE "\bU<n>\b" over
scenarios/ground -- exactly ONE U-item names itself. Five provisional
scenarios exist and four probably encode U-item defaults, but the mapping
is not written down, so it is not checkable. Same defect class as the
wrong premises, found inside the survey that proposed the fix. Now a
reported debt: open, lacking a reproduction: 9, target 0.
design.py carries the control design-baseline.py never had, asserted
directly: a row citing a nonexistent file must not count as reproduced,
using the exact path 2da19a4 deleted -- which the old tool called green.
design-baseline.py is marked superseded rather than deleted; it is the
evidence for how a wrong number got into a survey.
T06. The report is a FILE in ground-game under GROUND-WP-0002, committed
there, with a hub message that only points at it. It asks for no ruling:
it carries GR-E01's withdrawal, our reproduction debt, and two notes that
are explicitly not findings.
And it had to acknowledge something nobody anticipated. GROUND-WP-0002 is
finished -- all ten U-items were RULED 2026-08-03, every one confirmed as
the default we simulate, plus five of six provisional scenarios. The
survey said "0 of 10 ruled" two days later and this register was built
saying `reported`. That is the unread-inbox failure running in the
opposite direction: they answered and we did not collect it. The
instrument's first run surfaced it. They are `ruled`, not `applied` --
lifting the now-settled provisional flags is owed and is not done, and
make design shows them open until it is.
T07. evidence/CB-EV-0021. Two of six catches in this pass came from
execution rather than process (the role distinction from building it, the
ten uncollected rulings from running it), which is InnerLoop §Design
goal's prediction holding.
make self-tests, facts-check, loop-lint: clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Not a register; ADR-0012 D2 put that in GroundRules §Underdetermined. This
spec says what may go in it, what a reproduction must show, how a finding
dies, and how a trial game is run.
§1.2 is written against evidence rather than principle. A finding must
print the rows behind any number it claims, and the spec carries the table
of what shipped instead: a sum ("12 in the file"), a green scenario
("4/6/9 against 5/7/9"), and a condition named without checking which one
fired ("SOLVE on a face-down Problem"). "12" was arithmetically defensible
and still wrong about the game -- that sentence is the requirement.
§1.3's target is 0 reproductions that have gone green while open. GR-E01
would have tripped it four days before a human caught it by hand.
No baseline rate is quoted. The 33% was withdrawn by C2 and the first
honest denominator is T05's backfill; quoting a new number from a
discredited instrument is how the first one got in.
The trial protocol costs one flag: cb-play --record already writes a
finished game as a scenario, so a trial is that plus a sibling .md in the
player's own words. An observation is a NOTE until it has a reproduction,
and notes may not cross the repo boundary and expire at 30 days on the
existing provisional-age machinery. The maintainer's "I felt it was too
easy but then we lost" is the case the protocol is shaped around --
forcing it into a schema at the moment of observation would lose it.
loop-lint: no findings.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Implements ground-game's ruling of 2026-08-03. make all exits 0, 26
scenarios, rule coverage 59/59, and no scenario encoded the bug.
The rule ended up somewhere other than where I put it, and a gate moved
it. It went into legal_commands first; the AM-1 coverage gate then
demanded a scenario for the new GR-P05, and scenarios drive validate, not
the offer layer. A rule enforced only by the offer is enforced only for
clients that ask what is legal -- the browser would be filtered and a
scenario file would walk straight past it. Once GR-P05 moved into
validate, every condition in legal_commands was dead code, and the
layering test said so in those words.
And the reported case was not the one I reported. CB-WP-0018 and the
message to ground-game described SOLVE offered on a FACE-DOWN Problem.
Measured: validate already rejected face-down, so it never was offered.
Problem 1 is the Surface Problem, face-up from the deal -- the
maintainer's three inert SOLVEs were the HAND case, holding no Clarify
for a Clarify Problem. The ruling covers both so nothing is invalidated,
but the record was wrong.
Four conditions asserted separately, because one 'SOLVE is filtered' test
would pass with three of four implemented.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
T03: InnerLoop v1.7 plus loop-lint's own-cost check. Six passes
under-reported themselves by 30-45%, never once high, and the rule lived
only in evidence files having been re-derived three times. The READING
is load bearing, not the boundary: CB-WP-0018 T04 applied 're-run the
instrument at the moment of quoting' alone and its figure was correct.
So the operative instruction is re-run when you quote, and loop-lint
fails an evidence file naming its own workplan beside a dollar amount
without marking it provisional.
It binds forward from this pass. The check fires on seven historical
files which ARE the evidence for the rule; making them comply would edit
the record to remove the thing it proves -- the same category error as a
live fact: tag on a dated measurement, which this pass also hit.
Lifecycle, at the maintainer's instruction: ready -> active -> done,
where ready means declared and not started. loop-lint fails a workplan
that has started and still says ready, one that is active with
everything closed, and one that is done with an open task. The first
version of that check was WRONG and its own self-test caught it: it
stripped the leading status: assuming frontmatter, which silently
dropped a real task once the frontmatter said ready or active.
Both new checks then fired on this pass's own artifacts and both were
right to.
T04: CB-EV-0017. The new meta budget's first reading is a breach it
caused -- 27% against the 20% line, because this pass cost $31.18
against product passes averaging ~$21. Reported rather than exempted:
ADR-0006 D2 covers the instrument repairs but not the rule-writing, and
the honest reading is that this should have been two passes.
CB-WP-0018 settled at $36.53/95 against $28.08/82 last reported, 30%
higher. Seven for seven.
make all exits 0.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The two AM-4 budgets had the SAME scope -- one package, no dev edges --
while claiming to bound different things. AM-4b now measures the
workspace with dev edges: 57 crates / 725,258 lines where it read 29 /
317,021, having been blind to 28 crates and 408,237 lines, more source
than its own target.
Target 745,000, ~2.7% of room -- the same margin ADR-0008 D3 gave AM-4a,
applied to a number that grew because the instrument was repaired, not
because anything was added. The target moved to fit the measurement.
T02: proc-macros are COUNTED here and excluded from AM-4a, on purpose.
AM-4a asks what ships and a proc-macro never ships. AM-4b asks what is
acquired, and ADR-0007 D3's acquisition rule counts what the build
fetches -- 'it does not ship' is no answer to 'we downloaded it'. When
the rules disagree, the question each budget asks decides. Measured
share 109,585 lines / 15.1% against AM-4a's 36.2%, so ADR-0008 D2's
refusal to borrow the ratio was right by more than a factor of two.
Caught by this project's own earlier work twice: the mutation
find-string went stale and --self-test reported it BUILD-FREE (the check
CB-WP-0015 added after AM-4a's rotted for two passes), then the DFD gate
caught facts.toml carrying the old numbers.
CB-EV-0001 and ADR-0004 carried live fact: tags on historical readings.
A dated record asserting a CURRENT value is a category error, so those
occurrences are marked as-measured instead of retro-edited, and ADR-0004
gains a supersession note.
make all exits 0.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
InnerLoop v1.7. The purpose is written first and the number follows from
it: most spend on the task at hand, some on control, review and
improving the process. make status prints it above the figure, because a
threshold with no stated purpose is what let this number be argued three
times.
Soft 20% over a trailing 5, and the self-test enforces that the ratio and
the window are a PAIR: META_SOFT_PCT == 100 / TRAILING_PASSES. One meta
pass among n at parity cost reads 1/n, so 80/20 is one pass in five at
normal cost -- a five-pass window. The same 20% over three would have
silently also demanded the meta pass be half-price, which makes meta work
rushed rather than rare. Moving the ratio without the window goes red.
The phase setting is declared, argued and expiring in gates.toml, and
reverts on review_by unless re-argued. Verified live at 35%. One with no
reason or no expiry is refused rather than honoured, because a threshold
anyone may move is not a threshold.
Measured: the last five passes read 7% against the new line.
InnerLoop.md crossed the 400-line limit three times while this was
written and was fixed structurally each time -- the arithmetic, the
cost-per-response basis and the two review case studies moved to
InnerLoopReference.md. The limit was not raised.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
T03: input::describe writes a sentence per legal command; data-descs
carries them in step with data-targets; the ghost already following the
pointer shows the one for whatever legal target is under it, so the
explanation lands beside the target with no overlay layer to keep
aligned. ADR-0010 D1 binds -- the page renders it, never composes it.
Both mutations INITIALLY SURVIVED because the fixture's Attack card had
exactly one target, where an off-by-one shift and a truncation are both
no-ops. CB-EV-0014's lesson one level in: a fixture too thin to express
a failure is how the failure survives. Two attack targets now, both red.
T04: chaos rate d4 -> d8, window 2 open at 12 declarations, retiring if
an override changes nothing twice running. Window 1's condition was NOT
met -- both overrides changed the outcome -- so the mechanism is kept.
The weakest part of the decision is that it is a rate change argued from
n=2, so window 2 carries a falsifier: no override at all is evidence the
rate went too far, not that the mechanism is healthy.
InnerLoop.md hit 401 lines and the loadability gate fired; the rationale
moved to InnerLoopReference.md, structurally, per the standing precedent
that limits are not raised.
CB-WP-0017 settled at $9.48/40 against $5.19/23 reported mid-flight,
83% higher. Six for six, always low -- read by re-running the instrument
at the moment of quoting, which is CB-EV-0015's correction applied for
the first time.
make all exits 0.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Provenance (tier S, one paragraph in lieu of survey and ADR): the two
clauses mutation-check has reported inert since CB-WP-0005. AM-7's
scaling ratio was held up by a test literally named
replay_100k_events_is_linear_and_fast that computed both throughputs,
printed both, and never divided one by the other. AM-8's N=10 was held
up by a runner that does two.
Both are now red. AM-7 3/3, AM-8 2/2, M-D1-MUT 10/14, and ADR-0005's
>=10-of-14 prediction MET for the first time. Neither was closed by
amending the question away, which was the live risk: the denominator is
unchanged and the four unenforced rows are the four already
unenforceable.
AM-7 needed three estimators. Best-of-N per leg then divide (AM-6's,
correct for a floor on one number) gave 0.581-1.085 on an unchanged
binary; legs back-to-back gave medians 0.931-1.004; legs interleaved at
fold granularity give 0.987/0.991/0.989, and 0.989 under 8-way CPU
contention while absolute throughput fell 4x. The INDETERMINATE guard
demanded unanimity and failed a good measurement over one sample
0.001 under the floor; it now requires a two-thirds majority. The
control that matters: AM-6's constant-cost mutation halves throughput
and leaves this ratio at 0.999x green, so AM-7 is not a second AM-6.
AM-8 kept N=10 because the measurement said so. Perturbing the RNG only
from its fourth construction on: --runs 2 PASSES, --runs 10 fails. A
late-onset divergence is deterministic, not flaky, so it is a control
rather than a coin flip. Ten runs live on one scenario (make am8, ~2s)
rather than all 25 (47s a build). GameKernel 5b records it.
The full run also found AM-4a's own mutation stale since ADR-0008 D3
moved the target 250,000 -> 161,000 in CB-WP-0013 -- reported
HARNESS-BROKEN, no score published. The build-free half of that check
is now a --self-test assertion, so make all catches the next one.
mutation-check clauses may now carry their own verify and mutation, and
then the enforced flag is measured rather than declared; a declaration
disagreeing with its measurement is refused.
make all exits 0.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
ADR-0008, tier M (survey and ADR merged).
D1 — SH-3 retired as a gate, kept as a diagnostic. Investigating it
found a third defect, deeper than the two this pass was declared on.
Re-deriving batching from the raw transcripts, independently of cb-cost:
CB-WP-0011 pass 54 with tools 0 batched 0.0%
gap -> next decl 16 with tools 6 batched 37.5%
CB-WP-0012 pass 86 with tools 0 batched 0.0%
gap -> next decl 10 with tools 1 batched 10.0%
CB-WP-0013 so far 10 with tools 0 batched 0.0%
Zero batched turns in 150 in-pass responses; 37.5% in one gap, above the
20% floor. Batching needs two calls whose inputs are known at once —
orientation work. Implementation consumes each step's result before the
next. SH-3's window is since the last commit, which during a pass is
always implementation. The metric could not read above ~0% in the window
it was gated on. A floor the window structurally excludes is not a
target.
This pass's own declaration was also wrong: it claimed batching "has got
worse" (7.8-8.6% vs 1.1-6.3%). Differently-placed windows, not different
behaviour. Withdrawn — the same class of error, in the pass written to
correct it.
Not retargeting to match the measurement: the floor was not moved to 6%,
the gate was removed on an argument about what the quantity is worth.
The number is still reported; only the verdict is gone.
D2/D3 — AM-4a counts --edges normal,no-proc-macro: 157,202, not 246,250.
The target moves down with it, 250,000 -> 161,000, so the correction
hands back essentially nothing (headroom 3,750 -> 3,798). Three controls:
the exclusion drops exactly the five expected crates, only removes and
never adds, and is not a no-op.
The DFD gate then caught the follow-on it exists for — three historical
documents carrying live fact tags for a number that had changed. Not
rewritten; untagged, with a supersession banner.
AM-4b is deliberately not corrected: its proc-macro share is unmeasured.
gate-review now reads 0 due, 0 silent, 0 drifted — GATE-REVIEW earns its
first caught entry by forcing SH-3's re-justification, and the registry
has no silent gates left.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The metric six evidence files reported as 0.0% was never measured.
--shape-budget windows since the last commit, and it is read while
writing the evidence file — right after a commit, when the window holds
one or two responses. SH-1/SH-2 are location statistics and survive n=2.
SH-3 is a rate: at n=2 its only possible values are 0%, 50%, 100%.
cb-cost.py gains sh3_line(), which below a minimum sample prints
"insufficient sample" and no verdict. The floor is derived: if the true
rate were exactly the 20% target, P(zero batched in n) = 0.8^n, and at
n=14 that is 4.4% — so "0 batched in 14" rules out a target-meeting rate
at ~95%. Below that the tool has nothing to say and now says so.
The window was kept rather than split. SH-3 could have been given a
per-pass window, but the budget's purpose is the open remainder since
the last commit, and giving one of three metrics a different window
makes "the window" ambiguous in a tool three specs cite.
Four controls, three mutations, each red for its stated reason —
including the one the evidence files actually hit, where a refusal is
printed as a measured zero.
SessionShape.md §4 carries the correction with the real per-pass figures
(1.1%-6.3%), beside the eighth trusted-arithmetic instance. This is the
ninth, and the second in this same metric. It also shows what the frozen
0.0% hid: against the pinned 7.8%-8.6%, batching has got worse, and six
passes reported a breach that was moving underneath them.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
D2 the two exemptions — repairing the instrument that reports a breach
is always in budget, and the maintainer may authorize above-line work in
the workplan frontmatter. v1.5 had neither, and the first thing it did
was forbid fixing the budget's own defect.
D4 loop work cashes out into something runnable in the pass that invents
it, with the measured basis and the refutation condition stated.
D5 the tier table now triggers at M for changes to the loop's own
operation, closing the gap CB-RES-0005 §6 opened and this pass's own ADR
had to declare around.
The file went from 399 lines to 386: the chaos-roll calibration
rationale and the v1.0-v1.3 change log moved to InnerLoopReference.md.
Loadability is a gate too, and growing the file to fit a rule about
gates would have been the joke writing itself.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Scope cut first, on the maintainer's decision after a spend review: the
project is 38% product / 62% loop-meta, cost per response is 2.9x worse
than its best window, and INTENT stage 0 still lacks a CLI player and
bots. CB-WP-0005 and CB-WP-0006 cost ~$74 — 31% of all spend — for zero
measured efficiency gain. T02 and T04 are cancelled unstarted.
T01: SH-1/SH-2/SH-3 now report over the window since the last commit, and
the cumulative figure is retained but labelled "history, NOT the metric".
The prediction held decisively — window 655,744 mean context against
cumulative 255,307, a 2.6x gap against a 20% refutation threshold. A
cumulative mean over 1,094 responses cannot detect a worsening trend
because the history outvotes the present.
T03: `make shape-budget`, modelled on CB-01/CB-02. Soft thresholds are the
existing SessionShape targets; hard is 1.5x, set before the next
measurement per §Step 4. Deliberately not in `make all` — failing the
build on context would block committing, and committing is what closes the
attribution window and is the natural point to compact, so a gate that
blocks the remedy is a trap. It fires HARD on its first run: 656,574
against a 300,000 ceiling.
InnerLoop v1.5 establishes the soft 25% meta budget. Workplans declare
kind: product|meta|mixed and `make status` reports the share; mixed splits
50/50 and says so. Soft on purpose — a task already started may be
finished, because stopping mid-task to satisfy a ratio wastes the work.
What it forbids is opening new meta work above the line. A pass that
exceeds it must say so in its evidence and name the product work
displaced.
First reading: 68% OVER, of $74.22 attributed. Product reads $0.00 because
the only product workplan, CB-WP-0001, predates qualified task ids and its
bare T## labels collide across passes — stated in the output rather than
papered over.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Two rules, two different answers, which is the point of a task phrased
"implement, or amend and say why".
K18 is implemented. "Criterion benches driving the same scenario format at
scale" was false — the bench hardcoded its commands and never touched
ScenarioFile, while MetricsAndScenarios §3 pointed at a benchmarks/
directory containing only baselines/. benchmarks/synthetic-3p.yaml now
holds the workload and both the bench and bench_shape read it: the
workload is data, not code.
A second defect surfaced while fixing the first. After the bench switched
to the file, bench_shape still hardcoded the same sequence, so the
workload existed twice — deleting end_round from the YAML broke bench-test
while bench_shape kept passing. Duplicated-fact drift in executable form.
Both now read the same include_str! and deleting a command breaks both.
Explicitly not claimed: this does not unblock AM-3. AM-3's baseline is a
declarative game object — moves, turn order, rules. synthetic-3p.yaml is a
command list; the rules live in games/ground. Marking it as AM-3's subject
would compare a script to a game definition, which is the category error
AM-3 is blocked on. The file says so in its own header, where the next
person will be tempted.
K14 is amended. CommitWindow had zero non-test users and GROUND enforces
the same contract inline. Wiring GROUND through it was rejected: it would
change the serialized shape of `selections`, which four scenario files
assert by dot-path and every state hash depends on, for the sole benefit
of making a sentence literally true.
The deciding argument is INTENT's, not convenience: abstractions are
extracted from working games rather than invented in isolation, and no
concept becomes canonical until it survives a second concrete use.
CommitWindow was invented before any game needed it and has survived none.
Imposing it on GROUND would manufacture the first use rather than discover
it. So K14 states what is actually guaranteed, CommitWindow is marked
provisional in the source, and it carries a delete-by date of 2026-12-31.
Kernel spec->code link 16/18 -> 18/18, stated with the caveat the gate
prints every run: that is about names, not assertions.
Two self-tests broke and both broke correctly. rule-coverage's gate test
hardcoded "unlinked rules exist today" and failed when the last one was
linked; it now computes that and asserts the gate fails iff rules are
unlinked. facts' text check rejected k_unlinked once it became
legitimately empty; empty now renders as "(none)" and the check
distinguishes absent from empty.
M-D1-MUT: 8 of 14, unchanged — K14 and K18 are kernel rules, not
acceptance rows.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
INTENT design decision 8 of 10, unimplemented for six passes. cb-sim had
no flag parsing at all, so --replay had nowhere to go.
The bundle is manifest + commands.log + initial.snapshot + expected.yaml,
dev-only behind the scenarios feature and charged to AM-4b. The command
stream goes through the K11 framing built in T05, so a truncated bundle is
detected rather than replayed short — the two tasks compose rather than
duplicating.
The reviewer's D2 correction was real: this was not "a directory of four
files". Pass carried only the end state, RunOutcome::Failed was a
formatted String, and scenario.rs created an EventLog, appended to it and
never read it. All three had to change.
The first round trip failed to reproduce, and the cause is worth keeping:
state_hash_hex over a serde_json::Value is a different canonical form than
over the typed aggregate — Value's map is key-sorted, a struct serializes
in declaration order. The bundle was written with one basis and verified
with the other. A round trip written to recompute its own comparison value
would have PASSED this bug; it failed because the recorded hash came from
the producing process, which is control 2's entire purpose.
make replay-test implements ADR-0005 §6's four controls, 14/14: a
committed deliberately-failing fixture outside the corpus with covers: []
so it neither fails `make sim` nor inflates AM-1; a tampered recorded hash
must fail; a log short by one byte and a corrupted length prefix must be
rejected; and a mutated manifest seed must fail — which bites only because
replay re-derives the initial state from seed+setup and checks it against
the recorded snapshot, since restoring from the snapshot alone would leave
the seed inert. Plus a control on the controls: the bundle must still
replay after every mutation is reverted.
AM-7's hash-identical clause is re-earned. The probe records a hash per
per-game segment and replays each from its own genesis; folding from the
wrong seed now fails. That is the clause ADR-0005 §4 withdrew as
mutation-proven inert. The scaling >= 0.9x clause is still unenforced, so
AM-7 stays PARTIAL — reported, not rounded up.
Kernel coverage 15/18 -> 16/18. facts-check immediately caught the spec's
copy of that number going stale, on a number that moved the same hour.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
K11 is implemented: crates/cb-events/src/store.rs, magic + version header,
4-byte little-endian length prefix, append-only. Reimplemented not
assimilated per ADR-0005 §2 — no new dependency, and AM-4a/AM-4b are
unchanged at 246,250 / 317,021 because nothing entered the graph.
The operative clause is "detected", so corruption is tested rather than
assumed: a tail short by one byte, a half-written length prefix, a length
prefix corrupted to claim more than the file holds, foreign magic, and a
future format version are each rejected with a distinct error. A reader
that accepts a truncated tail is worse than no format, because it silently
returns a short history that looks complete.
AM-11 is earned. LogStore has two impls — MemLogStore and FileLogStore —
driven through ONE conformance(). The trait carries raw/set_raw precisely
so the corruption controls live in the shared suite: a format contract
that only one impl enforces is not a contract. The same shape is
retro-fitted to KernelRng, which is what AM-11 actually names: ChaChaRng
and NullRng now pass one suite asserting bounds, draw(1) == 0, determinism
across fresh instances, and shuffle preserving the multiset. They were
previously exercised by two separate tests, which is why "met, narrow" was
never earned and ADR-0005 §4 downgraded it.
K9 gets the assertion it did not have: snapshot at seq N + events N+1..M
must equal the from-genesis fold, hash-compared, on GroundState,
single-seed on purpose — AM-7's probe folds a multi-seed log, which is not
a replay of anything, and that defect is not repeated. Two positive
controls: the log must exceed 50 events, and the mid-log snapshot must
differ from the end state or "apply the remainder" is vacuous.
Proof it works: the exact mutation that SURVIVED in CB-WP-0005 — making
Snapshot::take discard its EventSeq — now fails on the K9 assertion.
AM-11's mutation breaks NullRng::draw to return its bound and the shared
suite fails. That is what M-D4-SWAP claims — either impl substitutable —
and exactly what two separate per-impl tests could never demonstrate.
M-D1-MUT: 7 -> 8 of 14. CB-EV-0001's scoreboard is refreshed: AM-2, AM-5
and AM-9 added, AM-6 moved to enforced, and the headline total corrected
from 4 to 8 — it had gone stale inside the same workplan that produced it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
AM-4c is withdrawn from the acceptance table and retained as a reported
diagnostic. GameKernel §5a carries the argument.
The ratio has no monotone better direction. INTENT's rule is "own the
semantics, assimilate the implementation": rising can mean owning
semantics properly or reimplementing what should have been assimilated;
falling can mean leverage or dependency bloat. A target requires knowing
which way is better. It is also redundant — AM-4a/AM-4b bound the
denominator and AM-2 bounds own-source density, so AM-4c is a ratio of two
already-targeted quantities.
Measured at withdrawal: 1,426 own lines per 100k third-party (shipped),
1,107 (dev). make dep-weight now prints both, labelled diagnostic — the
row was never actually reported before.
M-D1-MUT keeps AM-4c in its denominator on purpose and says so in the
output. Dropping it would move the score 7/14 -> 7/13 without enforcing
anything: a score improved by deleting the question.
Decided before Phase B deliberately, since ADR-0005 predicts own-source
growth that will move this ratio; deciding after would be the retarget
§Step 4 forbids.
A T01 correction found here. The AM-6 gate failed inside `make all` at
38,753 ev/s against 341,280 in isolation — a 9x drop, because cargo test
runs binaries and threads concurrently. A throughput assertion inside a
parallel harness measures contention, not throughput. T01's measurement
was valid; its gate placement was not.
Fixed by running it only where valid — #[ignore] plus `make am6` in
release with --test-threads=1, now 2.0M ev/s at 20.2x headroom — and not
by lowering the target, which T01 forbade. My first attempt did drift that
way, adding a debug "sanity floor" of 50,000, and was backed out: a second
threshold is still a second chance to tune.
The mutation then went SURVIVED on the first run after the move. 4,000
black_box iterations were calibrated against debug's 3.4x headroom and are
invisible against release's 20x. Raised to 100,000; back to red. A weak
mutation is not a fixed property of a row — it can become weak when the
row's measurement conditions change.
Tier S (amends one row, creates no capability), chaos d4=2, no override.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
T07 — CB-EV-0004. 70 responses, $16.03.
Test 1 MET: 1 spec -> 2, 58 rules -> 76, 1 source file -> 10, and the new
denominator came in at 83% naming K10/K14/K18. The stated failure mode — a
widened denominator reporting the same percentage — did not occur.
Test 2 UNMET: M-D1-MUT 4 of 14 against >=10, and wrong about what as well
as how much. The diagnosis was three absent kernel rules; the measurement
found eight rows with no instrument at all.
Test 3: quality held, and widening surfaced far more than the seven known
defects — six further uninstrumented rows, HDN #7 (rule-coverage's
self-test green while the tool was broken), a stale $248.46 invisible to
facts-check because it was untagged, and a fifth error class.
The clean test CB-WP-0004 was owed is now run, on a pass that used the
tools without building them. The two categories whose tools removed the
manual path are at 0 turns two passes on; the two that merely offered a
better option are now the entire mechanical cost of a pass. Absolute
mechanical cost fell $51.76 -> $6.13. The mechanism holds.
Reported because nothing else would: SH-3 batching is 0.0% this pass — 67
tool calls across 67 responses — against a 20% target, and mean context
315,170 against 200,000. SessionShape has stated these since CB-WP-0003
and none has ever been enforced.
T08 — the retrospective answers its question: yes, a weak mutation is the
new grep, and it is worse, because it fails in the opposite direction.
mutation-check's first run produced two SURVIVED verdicts and both were
the author's own no-op mutations.
The fifth error class: false accusation. HDN, TA, SSB and DFD all
under-report — a real problem passes. FA over-reports: it publishes the
claim that working code is broken, sends the next pass to fix something
that is not broken, and is more credible than the truth because it arrives
with a measurement attached. Thirteen instances, five classes, six passes,
and the newest class is one that hardening created.
One correction to CB-WP-0004 T06: "a gate only pays if it removes the
manual path" is a predictor of whether a gate saves money, not a criterion
for whether it is worth having. mutation-check fails that test and
produced the most valuable findings of the pass.
InnerLoop v1.4: where a claim rests on numbers, the adversarial reviewer
must read the assertion behind each quoted number and mutate it.
Re-running the command that prints a number is not verification of that
number. Second verification step to inherit the author's blindness; both
fixes replace re-derivation with adversarial execution.
CB-WP-0005 status -> done, 5 of 8 tasks, 3 cancelled into CB-WP-0006.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Four verdicts in evidence/CB-EV-0001 corrected in place with a dated
note, per ADR-0005 §4: AM-7 replay split (timing met, hash-identical
withdrawn), AM-10 withdrawn as written and restated as AM-10' (the K6
determinism lint it actually measured), AM-11 downgraded to unmet, and
AM-1b added to the scoreboard it was missing from.
The scoreboard gains an Enforced column carrying M-D1-MUT, because a row
can be measured and still enforce nothing and the table had no way to say
so. AM-6 now reads "met, 16.5x" alongside "not enforced — nothing
compares any number to 100,000".
A fifth correction surfaced that ADR-0005 did not list: AM-12 still read
$248.46, the figure CB-WP-0002 disproved and corrected to $93.15 four
workplans ago. It was stale in the evidence file ever since — untagged,
and therefore invisible to facts-check. Now tagged. A duplicated-fact
instance that survived the gate built to catch duplicated facts, because
that gate only checks copies that opted in. Recorded for T07.
GameKernel §5 carries the AM-10 withdrawal and AM-11 downgrade inline so
a reader of the spec cannot reach the old claim.
Phase C is deferred before starting, per the stop condition T02 wrote and
the maintainer's decision. It is scoped to five rules; the measurement
says eight acceptance rows have no instrument at all. Building it as
written would proceed on a diagnosis the instrument had just
contradicted. T04-T06 stay in the file with their analysis intact and
move to CB-WP-0006.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
AGGREGATE becomes a list of source roots and rule patterns become
per-spec, so the link runs over every numbered spec x every crate rather
than GroundRules.md x games/ground/src/lib.rs.
The prediction held on the first run:
AM-1b kernel spec->code link: 15/18 (83%) across 10 source files
unlinked: K10 K14 K18
Kernel rules are link-only by design, and the output says so: they are
kernel invariants with no aggregate, setup preset or command vocabulary,
so scenarios/kernel/*.yaml with covers: [K11] would be a tag in a
directory the runner cannot dispatch. Claiming scenario coverage for them
is the inflation this gate exists to prevent.
Per ADR-0005 §5 the kernel arm reports without feeding the exit code
until 2026-08-31, then binds — the date in the tool, not in prose, with
days remaining printed every run, because open-ended "gate it later" is
how AM-4's targets went unratified for four workplans. The self-test
asserts the gate returns 0 before that date and 2 after.
The zero-rules positive control is replicated on the new denominator: a
kernel regex that stops matching aborts rather than printing 0/0 as
though it were 100%.
The self-test passed while the tool was completely broken. A print(
inside say() became say(), so every real `make coverage` died with
RecursionError while --self-test reported all-ok — it only ever called
kernel_arm(quiet=True) and never executed the reporting path. The control
named the behaviour and did not assert it, which is precisely what this
workplan is about. Fixed by exercising the loud path and asserting it
prints, then verified by re-breaking say() and confirming both new checks
go red. Seventh instance of the harness-does-nothing shape, in the tool
written to find that shape.
Also caught by its own gate: a self-test label that printed "0 K-ids"
beside a passing ">5" assertion, because the detail string rebuilt the
pattern with different escaping. A label that contradicts its own check
is worse than no label.
k_rules, k_linked and k_unlinked are registered facts under facts-check.
A limit of that checker is recorded rather than patched: it is
line-based, so a tagged value that prose-wraps fails.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Duplicated-fact drift is the fourth error class and the only one with no
executable rule. No positive control catches it (both copies are
internally consistent) and re-derivation does not either (the copy
reproduces whatever it was copied from). It is caught only by reading a
copy against its source, which nothing in the loop required.
facts.toml holds 15 facts and is GENERATED by `make facts-gen` from
cb-cost, dep-weight and rule-coverage. The trap this task named — a
hand-maintained registry that becomes another drifting copy — is closed
by facts-check re-running the instruments and failing when the committed
registry disagrees with them. A stale registry cannot certify stale
artifacts.
An artifact quoting a fact tags it: **$93.15** <!-- fact --> with the key.
17 occurrences across 5 artifacts are now checked.
Falsified before being believed: changing CostAccounting.md line 158 from
$93.15 to $92.87 — the exact historical drift — produced exit 1 naming
the file, the line and the expected value. Tested against the class it
exists to catch, on a real artifact, not only in its self-test.
It then caught a live tag inside its own documentation example in
InnerLoop.md within the hour. Third time a gate has failed on its own
pass's work.
What it does not close is stated rather than implied: 22 untagged literal
copies remain and are reported, not failed. Tagging is opt-in, a number
can legitimately recur, and a gate that fires on coincidence gets routed
around. Naming the uncovered surface beats claiming the class is closed.
InnerLoop single-source-of-fact moves from prose to executable — v1.3.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Discharges the open item T07 raised. Values unchanged at 250,000 and
350,000; what was missing was a reviewed decision behind them, since
they had been set by the implementer in the commit that measured them
and that also changed the feature gating being measured.
The ADR supplies the argument T07's test requires -- why the targets
bind on FUTURE work rather than merely passing present work:
AM-4a leaves 3,750 lines of headroom (1.5%), about one small crate,
so any new shipped-runtime dependency breaches it almost immediately.
That is intended: the shipped runtime should be effectively frozen.
AM-4b leaves 32,979 lines (10.4%), deliberately looser -- dev tooling
should absorb one moderate dependency without a spec change, not two.
Both are ceilings on a quantity that only grows by choice. Nothing
drifts across them; only adding a dependency does.
Falsification condition stated: if a later pass raises AM-4a to
accommodate a dependency it wants, that is the failure the ceiling
exists to catch, and the answer is an ADR arguing for the dependency.
First ADR written under the correction/retarget test; sets the shape.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The retrospective question was whether the mechanism set is complete or
each pass still finds a new class. This pass produced both an eighth
instance AND a fourth class, so the answer is the uncomfortable one.
Ledger: 10 instances, 4 classes, across 3 workplans. Every pass has
produced at least one class the previous pass had not seen.
HDN harness-does-nothing 5 executable assertions
TA trusted arithmetic 3 re-derivation
SSB same-sample blind spot 1 assertions over ALL the data
DFD duplicated-fact drift 2 NEW -- reading a copy against source
DFD is genuinely distinct: no positive control catches it, because both
copies are internally consistent, and re-derivation does not either,
because the copy faithfully reproduces what it was copied from. Found
when an inlined price sheet went stale within an hour of T11 changing
the real one.
So v1.2 stops trying to enumerate classes in advance. Every error in
three passes was corrected in-session for under ~1% of the pass, so the
stated design goal is now cheap CORRECTION: keep raw data so numbers are
re-derivable, keep artifacts small and committed so a wrong number is
one grep from everywhere quoting it, give every number a command.
Plus the one rule the new class earns: single source of fact.
The original hypothesis is revised rather than confirmed. "A rule that
cannot be executed is not a rule" is wrong -- the two most valuable
corrections in the project came from a decorative rule that cannot be
automated (re-derive inherited numbers). An executable rule fires
reliably and catches one class; a decorative one fires unreliably and
can catch any class, including unnamed ones. Keep both.
Gates this pass: loop-lint caught 3 real violations on first run, then
failed on its own author within the hour when a T07 edit pushed
InnerLoop.md to 407 lines against its own 400 limit.
CB-WP-0003 complete: 11 of 11 tasks done.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Five scenarios encoded U-item defaults with no owner and no review date,
so they could shape the kernel indefinitely while looking handled.
Each now carries provisional_owner and provisional_raised, and the
runtime's ScenarioFile learned both fields (deny_unknown_fields meant
adding them to YAML alone would have failed every scenario -- the parser
had to agree).
make coverage reports every provisional item with its owner and age in
days, warns on any with no owner, and warns past 30 days. It WARNS
rather than breaking the build, on purpose: the ruling is ground-game's
to make and the kernel cannot make it for them. What the loop can
enforce is that evidence files list them, which is now stated in
GroundRules.
rule-coverage --self-test gained an assertion that every provisional
item has both fields, so the next one added without them fails a
command rather than passing quietly.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The copy in MetricsAndScenarios 1a went stale within hours of T11: it
still showed the sonnet intro rate as a comment after the real sheet
moved dated rates into data. A spec that duplicates a data file will
always drift from it; the copy is now a pointer that says so.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
make coverage now reports a second number: how many claimed rules are
also NAMED in the aggregate source. 58/58 tag coverage was weaker
evidence than it read as, and this says how much weaker.
Measured: 49 of 58. Nine rules are claimed by a scenario and appear
nowhere in games/ground/src/lib.rs --
GR-D07 GR-F02 GR-L03 GR-O03 GR-P01 GR-P02 GR-P03 GR-P04 GR-T01.
The gate REPORTS rather than fails, on purpose. Closing the gap by
adding those IDs to comments would satisfy the check without
establishing that any of the nine is implemented -- the overclaim
InnerLoop implementation rule 2 exists to prevent, and one CB-WP-0001
already committed once. Each needs its implementation confirmed before
it is tagged; promoting M-D1-LNK to a failing gate is correct after
that, not before.
Also added: a phantom check that fails when a rule id appears in code
that the spec does not define (currently zero).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The review step targeted the survey. Every serious error in this project
has been in measurement or build configuration, so the loop was
adversarially reviewing the artifact cheapest to fix and leaving
unreviewed the one where errors occur.
Step 2 now routes by risk: when the claim rests on numbers, the reviewer
gets the harness and the evidence file too, and must reproduce the
number independently rather than read about it.
The addition that matters more, because it was learned the hard way: a
reviewer re-derives the author's claims and therefore inherits the
author's SAMPLING. CB-WP-0002's dedup invariant was checked twice --
survey 206/206 groups, then the reviewer independently -- and both used
the main transcript. It is false in the 8-response subagent tree neither
looked at. Two independent verifications, one shared blind spot.
Rule: the reviewer re-derives on a different sample than the author
used, and where only one sample exists, says so rather than reporting
a clean verify.
Also recorded: what review demonstrably DOES do. $0.66 and $1.11 across
two passes, ~1% of each, both finding approval-blocking defects. Cost is
not a reason to skip it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A blanket "no retargeting in the measuring commit" rule would have been
wrong. CB-WP-0002 moved AC-1 three times in exactly that shape and every
move was correct -- each time a new instrument disproved the old figure.
Four legitimate corrections would have been forbidden to catch one bad
retarget.
The test is mechanical rather than a statement of intent:
correction the target moves and the implementation does not; legal in
the same commit provided the instrument's output is there
retarget the same commit changes both the target and the code the
target measures; requires an ADR stating why the new target
binds on future work
Applied retroactively: AM-4a/AM-4b are UNRATIFIED. They were set after
seeing the measurement, in the commit that produced it, with the
implementation changing too -- a retarget by this test. make dep-weight
is currently enforcing a target no reviewed decision stands behind.
Recorded as an open item; ratifying or changing them is a maintainer
decision, not an implementer's.
Also: specs/InnerLoop.md split into InnerLoop.md (process) and
InnerLoopReference.md (rubric, template, rules, definition of done).
Not a stylistic choice -- `make loop-lint` failed on the commit that
pushed the file to 407 lines against its own ~400 limit. The gate added
this morning to make that rule executable caught its own author within
the hour, which is the cheapest possible demonstration that it works.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The mechanism never fired. Two rolls across two workplans (9, then 2)
against ~0.2 expected firings at d10. At ~2 tier decisions per workplan
it would take roughly twenty workplans to see four overrides -- a rate
that prevents its own evaluation, which T06 named as the one option to
reject.
Raised to d4 for the next 12 tier declarations, then kept, reverted to
d10, or deleted on the evidence. Expected ~3 firings.
Cost stated up front rather than discovered: a chaos-L override on
S-shaped work buys a full survey + review + ADR, and the comparable
measured figure is CB-WP-0001 T03 at $9.91. The window should cost
$20-30. Cheaper than carrying an unevaluated ritual indefinitely.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The price sheet had two defects of one shape -- a schema that could not
hold the fact it needed, the same criticism the cost survey levelled at
the State Hub.
CA-16 time-boxed rates are DATA. Sonnet's intro price lived in a
`# intro ...` comment and was invisible to the collector that
reads the file. Now promo_input/promo_output/promo_until,
applied per response at its own timestamp.
CA-17 the 90-day staleness rule was prose in MetricsAndScenarios 1a
that every M-D2-CST verdict silently inherited. Now `recorded`
+ `max_age_days` in the sheet, and a stale sheet ABORTS.
Both are exercised by make cost-test: the promo rate must apply before
2026-08-31 and lapse after, and a 102-day-old sheet must trip.
Applying CA-16 moved AC-1 from $93.32 to $93.15 -- the $0.17 CB-EV-0002
predicted, now collected rather than noted. That is a legitimate
retarget under T07's distinction: the instrument disproved the target,
and its output is in this commit. The number has now been stated five
times ($248.46, $92.21, $92.87, $93.32, $93.15), each correction from a
different mechanism.
Evidence tables regenerated from the tool rather than hand-patched,
per CA-15 -- which is the rule that exists because hand-typed tables
were the only thing the adversarial review found wrong.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The 8k/10k per-task token budget was never referenced or enforced and
T08 blew past it silently. Replaced with a budget that can actually
fire.
The design constraint is the interesting part: per-task cost needs the
commit that CLOSES the task, so a per-task budget is unavoidably
retrospective -- it can only report a breach after the money is spent,
which is the dead-policy failure again. What IS observable mid-task is
spend since the last commit, because the transcript is append-live. So
the budget binds on the open remainder.
CB-01 budget = USD since the last commit, via `make cost-budget`
CB-02 soft $10.00 (state progress, decide), hard $22.00 (stop)
Calibrated on the 32 non-empty commit intervals of CB-WP-0001: p50
$1.40, p90 $9.36, max $10.80. Soft sits just below the observed maximum
-- it would have fired exactly once on the calibration pass. Hard is ~2x
the observed max, a value never reached in 32 intervals, so reaching it
means the session is doing something the data has no example of.
Both thresholds are set ABOVE every observed value, so they bind on
future work rather than ratifying present work -- the distinction T07
is about.
Stated limit: it is a command, not a daemon. An agent that never runs it
gets no signal, which is the dead-policy failure one level up. Mitigated
only by being free to run and on the one command surface.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The task's original premise was wrong and is recorded rather than
deleted. It was written to prescribe one task per session; measurement
says the variable is context, not turn count.
Measured:
- compaction cut context 27x (542,991 -> 19,974) and cost/turn 3.1x;
the 202 turns after C1 cost less than half the 136 before it
- a turn costs $0.010 at 20k context and $0.270 at 540k
- break-even for a compaction is 2-11 turns, so: compact whenever
context exceeds ~300k and work remains
- a fresh session is NOT free -- cold start floors at ~51k and must
then re-read the artifacts a compact summary already holds (~66k).
Prefer compaction to continue work; prefer a fresh session when the
task changes, because then prior context is pure overhead.
cb-cost now emits SH-1..SH-3 so the targets come from the instrument
rather than from analysis, per InnerLoop v1.1. All three are UNMET
(mean context 232,982 vs 200,000; p90 492,042 vs 300,000; batching 7.8%
vs 20%) and are reported unmet rather than retargeted -- retargeting in
the commit that first measures is precisely what T07 exists to prevent.
Eighth error instance found while writing this: CB-WP-0001's claim that
"0 of 330 tool calls were batched" is wrong. 330 was the count of
single-call responses, not the total; 31 responses batched, covering 76
calls. It was carried into this workplan unverified. Trusted-arithmetic
class -- the one the T01 audit flagged as having no executable defence,
confirming that finding within hours of making it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
41 rules classified executable / checkable / decorative, each tagged with
the failure class it catches. Counts: 11 executable, 22 checkable, 4
decorative (one of them dead policy).
Audit: history/260731-inner-loop-rule-audit.md
New tools/loop-lint.py makes 7 rules executable (tier declared, chaos
roll recorded, tier-L review trail, unmeasured-in-evidence, whole-file
loadability, reporting tools expose --self-test). It found three real
violations on its first run, none previously visible:
- specs/ArchitectureBlueprint.md was 543 lines against a ~400 limit
the loop has stated since v0.2 and never measured. Split at its own
section boundaries into Blueprint (1-8) + Runtime (9-15).
- tools/dep-weight.py and tools/rule-coverage.py had positive-control
logic and no --self-test, so nothing verified the control worked.
Adding rule-coverage's self-test exposed a latent instance of the exact
class this workplan is about: if the spec regex stopped matching, rules
was empty, missing was empty, and the tool exited 0 reporting "0/0" --
a silent pass, in the tool that reports our headline AM-1 number. Both
tools now assert they found something before reporting.
Two demotions applied in the spec rather than left implicit: "structured
over prose" is marked guidance (nothing can check it), and the 8k/10k
token budget is struck through and marked DEAD POLICY pointing at T05.
The audit's uncomfortable finding: rule 13 (re-derive inherited numbers)
has no mechanical form, is deliberately left decorative, and caught the
LARGEST error in CB-WP-0002. That is a counter-example to this
workplan's own hypothesis. "A rule that cannot be executed is not a
rule" is wrong as stated; the defensible version is that such a rule
cannot be relied on to fire, so it must not be the only defence for a
class that matters.
Class coverage: harness-does-nothing has five executable rules;
trusted-arithmetic has ZERO and produced the largest single error.
make loop-lint and make self-tests wired into `make all` and CI.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Answers the question the workplan posed: did v1.0's "every metric names
its instrument" rule stop a metric being written without a working
instrument? No. AC-1 named `cb-cost --pin` before that tool existed and
set a hand-computed target of $92.87; the tool returned $93.32. The rule
was satisfied completely and the metric was still wrong.
v1.1 adds:
- the instrument must exist and the target must come out of it;
targets are provisional until the tool emits them
- a number inherited from earlier work is re-derived before use as a
target, or cited as unverified
- every reporting tool exposes --self-test, run before the number
- cost is in the definition of done; M-D2-CST may not be uncomputable
The cost of CB-WP-0001 was stated four times before it was right --
$248.46, $92.21, $92.87, $93.32 -- and each correction came from a
different mechanism: re-derivation, adversarial review, and the positive
control. None found more than one. That is the case for keeping all
three.
CB-WP-0003 T10 predicted the next error would be harness-does-nothing.
It was not, twice. Trusted arithmetic over real data would pass a
positive control; and a property verified on 206/206 groups of the main
transcript is false in the 8-response subagent tree that neither the
survey nor the reviewer examined separately. Review structurally cannot
catch the second -- re-deriving on the same sample reproduces the same
blind spot.
Workplan status: done.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- InnerLoop definition-of-done gains a cost row: M-D2-CST may no longer
be recorded uncomputable, and composition must be reported, not only a
total.
- InnerLoop Step 5 gains the --self-test contract: every tool that
reports a number exposes one, and it runs before the number does.
Rationale attached, because the case that motivated it is the one
review cannot catch — survey and reviewer both verified the same large
sample and both missed the small one.
- make cost / cost-test / cost-pin on the one command surface; cost-test
in `make all` and in CI.
- Hub now holds the measured figure for CB-WP-0001: 80.6M in / 323.6k
out against the 401,100 it previously estimated, low by ~200x. The
event states plainly that the hub schema cannot represent the 88% of
cost that is cache, and names `make cost-pin` as the authority.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Collector per ADR-0003: enumerates every transcript including the
subagents/ tree, dedups by requestId, prices per model and per cache
TTL, attributes on (prev_commit, this_commit] intervals, and reconciles
to the cent or aborts.
The positive control caught a real defect on its very first run against
real data, which is the entire argument for writing it:
CA-02 assumed usage is identical across the lines of one requestId.
True in the main transcript (206/206 groups, verified twice — by the
survey and by the adversarial reviewer). FALSE in the subagents/
tree, where output_tokens is a running count: one response reads
5, 5, 195 across its three lines. First-wins scored it at 5.
So CA-02 now splits: input-side counters are charged once and must be
identical (assertion retained); output_tokens resolves to the max
(CA-02a). AC-9 pins the exact 5,5,195 case as a regression test.
The acceptance target moved again as a result, for the third time:
$248.46 -> $92.21 -> $92.87 -> $93.32. AC-1 was computed by the same
first-wins method the tool just disproved, so the tool failing its
target was the tool being correct. Target updated, not the tool.
Measured at the pin: $92.21 main + $1.11 subagent = $93.32, residual
$0.000000. 32.5% UNATTRIBUTED, reported as its own line per CA-10.
make cost / cost-test / cost-pin wired; cost-test added to `make all`
and to CI, where it gates the collector's assertions without needing
transcripts present.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Normative spec for the cost model (dedup by requestId, per-model and
per-TTL pricing), the attribution contract (ending-at-commit intervals
scoped by sessionId), and the reported shape (composition, not a total).
Every acceptance row names the command that produces its number, per
InnerLoop Step 4. AC-5..AC-8 are the positive control: cb-cost
--self-test must fail on a dedup violation, on zero responses, on a
missing subagent tree, and on 5m cache priced at the 1h rate.
The feasibility check earns its place here: AC-1's $92.87 is reachable
only if CA-06 holds. The earlier $92.21 target was reachable only by a
collector with the exact blind spot the survey documented.
MetricsAndScenarios 1a superseded. Two of its rules struck through
rather than deleted, because both were wrong in instructive ways: "if
the cache split is unknown, count all input at full price" would have
priced 80.5M cache reads at 10x, and "the hub already records tokens"
named as a source what is only ever a lossy sink. M-D2-TOK demoted --
tokens are not comparable across models or cache states.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Adopts both remediations from CB-EV-0001 §4 (maintainer decision).
Option A — serde_yaml is now optional behind cb-game-runtime's
`scenarios` feature. The scenario module, the ScenarioGame impl and the
string parsers behind it are cfg-gated; cb-sim opts in explicitly. Both
configurations compile and lint clean under -D warnings.
A trap worth recording: `default-features = false` on a *member*
dependency is silently ignored when the workspace dependency does not
specify it. The first attempt gated nothing while looking correct — the
build succeeded and cargo tree still showed all six YAML crates. Fixed
by setting it on the workspace dependency. This is the positive-control
failure mode in miniature: success was not evidence the change applied.
Retarget — AM-4 now measures third-party source under audit, split by
build configuration, replacing a crate count that was unreachable
without undoing K5/K7 and that does not compare across ecosystems.
Re-measured via the new `make dep-weight`, whose own positive control
refuses to report when any crate's source cannot be located:
shipped runtime 23 crates 246,250 lines target <=250,000 met
dev toolchain 29 crates 317,021 lines target <=350,000 met
own source 3,408 lines
Scenario tooling costs 70,771 lines a shipped game never compiles —
the split the single number was hiding.
Targets are set at current measurement plus headroom, so they bind on
future growth rather than retroactively passing what had failed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The loop has now survived CB-WP-0001 end to end and is revised from what
that pass actually taught, recorded in
history/260731-inner-loop-retrospective.md.
Held up: the adversarial review (one blocking and three significant
findings on its single use, resolved with new measurements rather than
argument), the parity-cap rule, the provisional U-item mechanism, and
the ADR gate.
The gap: both serious errors in the pass were measurement errors, and
the loop caught neither. Review reads prose; these were claims about
numbers. In both cases a harness ran successfully while doing no work —
a JS harness timing rejected no-ops, and a Rust benchmark computing
throughput for rounds that never completed.
v1.0 adds:
- Measurement validity: a harness must assert it performed the work it
reports. A number from a run that cannot prove it did the work is void.
- Metric feasibility: every metric names its instrument and is checked
reachable against the contracts in its own spec.
- No silently-ignored input; decisions get commands, not defaults;
scaffolds are exercised or marked; coverage gates that count tags say
so.
- Evidence must state what a comparison does not support.
- The chaos roll is recorded even when it changes nothing, so a
mechanism that never fires is visible rather than assumed.
CB-WP-0001 is complete: 9/9 tasks done.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
