Why capability has not moved despite everything being implemented, answered by
measurement rather than opinion, at 0e1be8be:
F-A The serving path is 100% symbolic. reader.py (490), proof_chain (3,916),
both surface composers: zero geometric references. All served capability
comes from a template parser feeding an ROBDD engine.
F-B The reader INVERTS its own ratified design. ADR-0252 §2 stage 1
prescribes containment/transfer/accumulation/comparison and bans
surface-slot regexes verbatim; the implementation is surface-slot
regexes over categorical/member/ordering/propositional. Corpus agrees:
1.28% read rate, possession/transfer/quantity absent, if/then (0.1% of
sentences) holding 21% of the inventory.
F-C Template architecture cannot produce chunk lifts — each construction is
one hand-fitted rule; generalization is zero by construction. This is
why lifts have only ever arrived as overfit slivers.
F-D Geometry does exactly two cognitive jobs and both are stranded:
salience->attention (bypassed by every licensed lane) and
relation_compiler's Hamiltonian ground-state solver (off-serving,
behind a 1.0% math reader). Real geometric computation exists — it
"NEVER evaluates scale*input+offset in Python" — and nothing feeds it.
F-E The §5 NO-GO killed alignment-on-points-under-similarity. The
composition branch — relations as OPERATORS, meaning as composition,
the pattern relation_compiler already embodies — was never tested.
Plan: docs/plans/2026-07-28-perception-arc.md. Phase 0 pins the 1.28%/1.0%
read-rate floors as the metric the arc is accountable to. Phase 1 is a
§5-shaped pre-registered GO/NO-GO on hypothesis H2 (compositional
core-primitive reading over operator encodings; NO-GO branch declared: a
principled compositional grammar instead — either verdict exits the template
treadmill). Phase 2 builds ONE perception layer as combinator core + declared
construction lexicon, so coverage growth is table rows, not parsers. Phase 3
reconnects the Hamiltonian corridor to serving through the now-honest licence
machinery. Phase 4 deletes what earned it: G-3's template-widening program
superseded, curriculum-serve machinery parked. Phase 5 prepares the ADR-0252
amendment and the use_salience ruling for one-word decisions.
Explicitly untouched: core/physics, the algebra, the reliability gate, the
ledger discipline — correct and load-bearing. The enumerative reader stays
serving until its replacement proves out: supersession-in-place, per the ADR's
own preserved rule.
G-24 registered Tier A. G-3 as previously specified is superseded, recorded in
the execution plan's closing addendum rather than silently.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Intended as a pull-request body. Neither remote could host one, so it lives in
the repository instead — which is the more durable home anyway, since it travels
with the commits it describes rather than sitting in a forge that may be
re-synced.
- GitHub: the App installation lacks pull_requests:write. POST /pulls returns
403 "Resource not accessible by integration".
- Forgejo: the agent proxy denies core-gitquarters.acbcontent.org at CONNECT
(403, recorded in the proxy's own relay-failure log). Organization network
policy, not a transient failure.
Indexes all 29 commits from 797ebad5 (exclusive) to 199dc72d, each with its
diffstat and an explicit revert classification — Independent, Coupled (with the
group named), or behaviour-changing with the consequence stated. Plus
cherry-pick groupings, three named rollback targets, and the arc's three results.
Records the mirror hazard that already fired once: GitHub's main was reset
backwards from 199dc72d to 797ebad5, which is EXACTLY Forgejo's main. Per
AGENTS.md:280 GitHub is a mirror only, so this was almost certainly a re-sync
from the canonical remote rather than a rewrite. The local clone still had
everything and restored GitHub by fast-forward. Until Forgejo receives this arc,
any further mirror sync will revert GitHub again — which is the single most
important operational fact in this file.
Also records the three hollow mechanisms found in my own work during the arc,
and the two earlier statements that were wrong and were corrected in place
rather than overwritten.
No runtime behavior. One new file.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
The last docket item. Three decisions that cost nothing to make and something
real to leave silent, plus one proposal withdrawn because instrumentation showed
it could not be built as specified.
docs/specs/postures.md. Each posture names the CRITERION that would change it,
because an unstated boundary reads as an unexamined one and a future reader
cannot tell a deliberate limit from an accident.
P-1 (R-1 A, G-12) — efferent action deferred, in scope eventually. No efferent
surface beyond typed trace-folded tool operators until (i) the chooser exists
and is governed AND (ii) an efferent falsification bench exists. Both halves are
load-bearing: without (i) CORE acts with no account of what it should do next;
without (ii) it acts with no way to be shown wrong. ADR-0211's bench-level
prohibition is narrower than this and remains in force.
P-2 (R-5 A, G-11) — identity enforcement stays scoring-only until a named
held-out benign/adversarial corpus shows separation on the certified metric at a
floor named BEFORE the run. Same pre-registration discipline that made the §5
NO-GO full credit, for the same reason: a refusal gate authorized on a floor
chosen after seeing the numbers is not evidence, and identity refusal is
expensive to get wrong in both directions.
P-3 (R-6 A, G-17) — non-text ingest deferred, with the falsification bench as
the standard. A modality enters serving on the SAME terms text did: named
held-out corpus, holds/bites predicates, wrong=0-or-refuse. No modality is
admitted because the substrate can represent it. That makes the 59 sensorium
modules a capability awaiting evidence rather than an unexplained absence.
P-4 (R-11 B -> second ruling) — THE INTERIM FABRICATION GATE IS WITHDRAWN.
R-11 ruled "measure first, then re-ask". The measurement ran over 11,199
distinct serving-path inputs and 23,562 clauses and returned ZERO outside the
verified inventory — which is not a clean bill of health. It has a cause:
atom_fact's template is "{p}" — one slot, no literal anchor. It matches ANY
string, and it is one of the 19 verified constructions.
So "outside the verified inventory" is NOT A WELL-DEFINED PROPERTY. Nothing is
outside it. The proposed gate would have refused nothing while presenting as a
safety mechanism — a mechanism whose failure state is indistinguishable from its
success state, which is this repository's dominant defect class.
This reframes G-2, and the reframing is the finding. "every dog is a mammal" ->
member(every_dog, mammal) is not the reader ADMITTING an out-of-inventory
construction. The surface is admissible; the reader assigns it the WRONG
RELATION. The defect is in the mapping, not the admissibility set, and no gate
over the admissibility set can catch it.
Second ruling, delegated: option A is WITHDRAWN as unimplementable as specified,
NOT deferred — a deferred option is one that could be built later, and this one
cannot be built at all against the inventory as it stands. Option C is
operative. The fabrication ADR inherits the boundary question: either atom_fact
is narrowed so admissibility is decidable, or the guarantee moves from
admissibility to MAPPING CORRECTNESS. The evidence points at the second.
METHOD NOTE, recorded because the number nearly shipped wrong. Two earlier
passes were both invalid and both looked fine. The first matched templates
against whole multi-sentence inputs and reported 100% out-of-inventory when
every clause was in-inventory. The second misread the sentence-splitter's tuple
and measured "." 23,562 times. The catch-all was found only by a NON-VACUITY
CHECK — asserting the matcher could still say no to "most birds can fly" — and
it could not. A measurement that cannot fail is not a measurement, and that is
the same standard this arc applied to every pin it shipped.
Closes G-11, G-12, G-17. The entire adopted docket is now executed:
R-7 -> R-12 -> R-3+R-4 -> R-9+R-2 -> R-13 -> R-8 -> R-1/R-5/R-6/R-11.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
The reachability gap turns by PROMOTION for the first time: 15 -> 14. PR-3b's
earlier turn was by deletion, which is free. This one costs 22.1s on every push,
which is exactly why it is a decision rather than a default.
THE RECOMMENDATION WAS RE-DERIVED, NOT EXECUTED — and it did not survive as
written. PR-4b recommended `teaching` on the strength of G-22:
test_ratification_ceremony.py was red on main, in `teaching`, run by no gate.
That hole was already closed a different way — G-22's fix promoted that single
file into smoke. So `teaching` would have added 9 net-new files, NONE with a
demonstrated escape, and promoting on a justification that has already been
spent is precisely the failure this arc exists to close.
Promoted on a stronger, doctrinal basis instead.
tests/test_epistemic_invariants.py enforces AGENTS.md's "Teaching and mutation
safety" non-negotiables in substance — INV-22/INV-23: an unmarked proposal and
an unmarked reviewed example default to SPECULATIVE, the status enum has exactly
its four positions, a pack-mutation proposal carries no hardening flag. Durable
corpus mutation is the one irreversible thing in this system, and the suite
guarding it ran on no pre-merge gate. That justification does not require an
incident to have happened yet, which makes it the better one.
Measured: 22.1s / 122 tests across 9 net-new files (the tenth was already on the
gate). Both invocation sites updated, GATE_SUITES extended, `teaching` removed
from UNREACHABLE_BASELINE in the same edit, ratchet lowered <= 15 to <= 14 —
the both-directions discipline that has caught this author twice.
A HOLLOW SPOT IN PIN 3 ITSELF, found by sabotage-testing the promotion rather
than trusting it. _suites_invoked_by_gate_tiers() takes the UNION of
scripts/hooks/pre-push and scripts/ci/local-ci.sh, so a suite present in either
satisfies it. Removing `teaching` from local-ci.sh alone PASSED SILENTLY: the
two scripts could drift from each other invisibly while local-ci.sh states in
prose that it runs "the same four steps as scripts/hooks/pre-push". A developer
running one would get coverage the other advertises and does not deliver.
Closed by test_the_two_gate_scripts_invoke_the_same_suites, scoped to the
numbered (n/m) gate steps — a discriminator both files already use — so it reads
the gate tier without parsing shell control flow. The identical sabotage now
fires. Verified in three directions: removed from one script (fires, and did not
before), removed from both (the pre-existing assertion fires), restored (green).
THIS IS NOT THE CI-PARITY PIN N-9 KILLED, and the distinction is recorded at the
assertion site so it cannot be re-litigated a third time. That one asserted
TEST_SUITES["smoke"] == smoke.yml and was reverted in 50fa287d because
AGENTS.md:280 makes GitHub Actions billing-locked dead signals — parity with a
dead signal buys nothing. Both scripts here are LIVE local merge bars.
Gate is now four steps: smoke + warmed_session + deductive + teaching.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Rank 6, the last docket item before the posture statements, and the throughput
frontier N-5 called "one ruling wide". The ruling landed. It closed the frontier
rather than opening it, and that is the correct outcome.
R-8 ruled C: committing to an entailment and correctly declining to commit are
DIFFERENT capabilities, licensed on DIFFERENT evidence, and may not be pooled.
THE FINDING: split, ZERO bands license — not four.
band unknown floor entailed floor
philosophy_theology_contrast 652 0.989925 8 0.000000
philosophy_theology_modal 652 0.989925 8 0.000000
physics_causal 653 0.989940 7 0.000000
systems_software_causal 653 0.989940 7 0.000000
(pooled, the old basis) 660 0.990046 -- clears by 0.000046
N-5 recorded that "four bands would earn SERVE the moment a ledger is sealed."
That is true on the POOLED basis and false on the ruled one. Those four cleared
theta_SERVE=0.99 only because 7-8 entailments were counted alongside 652-653
correct refusals to reach 660. The licence was manufactured by the pooling, not
earned by the evidence. conservative_floor(9,9) is 0.000000 — the Wilson lower
bound at nine trials is not merely below theta, it is zero.
This is exactly why the plan blocked PR-14 on R-8 instead of sealing first and
ruling after. A ledger sealed on the pooled basis would have granted four
licences that the mix rule then had to revoke — and revoking a granted licence
is the expensive direction, as PR-12 had just demonstrated across 21 deduction
bands in this same session.
DELEGATED RULING — R-8 C's entailed floor N: no new constant.
The packet recommended "C, with a floor from A applied to the entailed
capability only," and N was never named. It does not need naming. theta_SERVE
=0.99 through conservative_floor — the bar every other capability already meets,
657 distinct correct decisions — applied to each separated capability on its own
evidence licenses nothing, by a factor of 73. Inventing a second, weaker
constant for the capability that most needs the strong one would institutionalise
two standards, which is precisely the objection that sank option B. Recorded
under the standing delegation with its reasoning, so it can be overturned.
NO LEDGER IS SEALED, AND THAT IS THE DELIVERABLE. Under the rule there is
nothing to license, and the registered missing_ok=True absence already says so
once. An artifact whose only content is its own emptiness would be a second
statement of the same fact — the defect class registered as G-23 this same week.
THE USEFUL RESULT IS THE SHAPE OF THE GAP, WHICH POOLING HAD HIDDEN.
Non-commitment serving is FOUR TO FIVE distinct query atoms short per band.
Entailed serving is ~648 short. Those are content tasks of completely different
size, and a single pooled figure could not distinguish "almost there" from "two
orders of magnitude away". Four cases is a morning's work; 648 is a program.
Nobody could see that before the split.
Delivered:
- curriculum_serve_entailed registered in CAPABILITY_LEDGERS, missing_ok=True
(absent = nothing licensed, the honest state), with the rule declared in the
manifest table rather than at a call site (ADR-0263 rule 5)
- an audit source for it — demanded immediately by the manifest's own pin
(test_every_licensed_capability_has_an_audit_source went red the moment the
capability was declared, which is PR-5's declared-table discipline working)
- ADR-0264 §5 amendment recording the ruling and correcting §4.1's "four bands
would earn SERVE" expectation. Changes no decision in that ADR;
curriculum_serving_enabled stays False
- tests/test_curriculum_outcome_mix.py on the gate: both capabilities pinned,
the shortfall pinned exactly, AND the counterfactual pinned — pooling
licenses 4 — so the argument for the rule cannot drift away from the number
it rests on
The serving path needs no change, and that is stated rather than left implicit:
under the rule nothing is licensed, so there is no licensed-vs-disclosed branch
to route. Building that machinery now would be building for a case that cannot
occur yet.
Closes G-10.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Rank 5 on the docket, and the first item this arc whose expected outcome is
LOSING capability.
conservative_floor is a one-sided Wilson lower bound, and Wilson assumes
independent trials. CORE's pipeline is deterministic, so replaying an identical
case is not a second trial — it is the same trial with a guaranteed outcome.
The deduction sealer folded the raw practice corpus and counted every replay,
so bands recorded 720/720 committed whether that was 720 decisions or 28
decisions seen 26 times each. conservative_floor(720,720)=0.990868 cleared
theta_SERVE=0.99; conservative_floor(28,28)=0.808413 is not close.
RESULT: 25 licensed bands -> 4. Twenty-one demoted.
Survivors: en_conditional_chain, en_disjunctive, en_verb_fact,
en_verb_universal — the only bands whose corpus holds enough independent cases.
What the demotion did and did not do, stated exactly. NO ANSWER CHANGED. No
answer became wrong. `wrong` stayed 0 across all 25 bands. The engine is
exactly as correct as it was. What changed is the CLAIM attached to the answer:
21 bands now serve the same sound conclusion prefixed with "(reasoned, but I
haven't yet earned a verified track record on arguments of this shape)". The
reasoning did not get worse; the boast did.
The producer is hardened, which is the half that stops this recurring.
seal_ledger now refuses outright — assert_sealed_evidence_distinct compares the
ledger about to be written against the corpus's distinct-case count and raises
BEFORE anything is written. The curriculum sealer has carried this guarantee
since ADR-0264 R9; the deduction sealer had no equivalent, which is exactly how
the exposure arose. Catching it at seal time rather than in an audit matters:
an audit finds a padded ledger after it is committed, trusted, and gating a
live flag, and unwinding that then needs a ruling — it needed one.
A HOLLOW GUARD OF MY OWN, caught by sabotage rather than trusted. The first
version compared all_gold_problems() with distinct_gold_problems() — two
functions that agree with each other by construction, and neither of which is
what build_ledger folds. Reverting build_ledger to the raw corpus walked
straight past it AND re-sealed the inflated artifact. Same defect class the
ledger already suffered: a guard checking something adjacent to the thing it
protects. It now takes the BUILT LEDGER as input, because the invariant is
about the artifact. Re-running the identical attack: refused, wrote nothing.
Pinned as test_sealer_refuses_a_ledger_that_counts_replays.
A SECOND MISS, caught by the gate rather than by me. My blast-radius scan
grepped for deduction_serve_license / deduction_serve_ledger / 720 and found
nine files. It missed tests/test_construction_inventory.py, which asserts
deduction SURFACE STRINGS and never mentions the ledger — so a pattern search
over the wrong noun could not see it. The full gate found it. Recorded because
"I scanned for the blast radius" is worth exactly as much as the scan's key,
and mine was the wrong key.
That test is the #138 fabrication defect pin, and it is updated WITHOUT being
weakened. Both surfaces now carry the disclosure prefix, and the fabrication
survives untouched — which is a sharper finding than before: a disclosure
prefix looks like it might cover the hazard, and it does not. The user is still
told they said "furthermore". Exact-match assertions kept on the full strings
so neither the fabrication nor the licence state can drift unnoticed, plus an
isolating assertion that the only difference between clean and leaked recitals
is the fabricated premise.
Four records corrected in the same PR, each of which would otherwise have gone
on asserting capability that no longer exists:
- CAPABILITY_LEDGERS' manifest note ("25 bands at 720/720 wrong=0")
- the published CLAIMS.md claim — caught immediately by G-22's own pin, which
is that fix earning its keep three days after landing
- test_volume_honesty.py's module docstring, which said the exposure was
"NOT silently repaired ... changing it is Shay's ratification, not a
test's". R-13 IS that ratification, so the line went stale by being acted on
- the exposure inventory's framing: it now pins the repair, not the gap
test_volume_honesty's sealed-vs-producer assertion is strictly stronger than it
was. It compared sealed committed counts to the producer's committed counts and
both were the inflated 720 — the assertion held perfectly while the number it
agreed on was the wrong number. Two records agreeing is not evidence that
either is right. It now pins committed == distinct, plus a non-vacuity check.
Nine tests encoded the unearned expectation and were updated to the honest one.
Closes G-19 and H-1.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Rank 4 on the docket. The L10 always-on lane has passed a 5000-beat soak since
2026-07-19. That result lived as PROSE IN A CONTRACT FILE — no machine-readable
artifact, no run SHA, no rerun path, and a deterministic_digest that report.py
computed and nothing pinned. The contract's own closing line asked for the pin
("Pin it once the lane is trusted so a regression flips it") and it was never
added. A recorded prose result with no re-derivable artifact is testimony, not
evidence, and that file is exactly where the distinction was meant to be
enforced.
R-2 first, before the run — as the plan required, so the soak measured a
decided question rather than deciding one by accident.
R-2 was ruled B (beats govern cognition; wall-clock permitted in telemetry
only). Measured: this lane carries NO WALL-CLOCK AT ALL — runner.py,
predicates.py and report.py contain no time, datetime, or timestamp of any
kind. That is stronger than B requires, and it is KEPT rather than "completed",
for three reasons now recorded in the contract:
1. B's permissive half is a permission, not an obligation. Adding a field so
the code matches a ruling's letter, then adding a pin to guard the field,
is a mechanism whose only purpose is to be guarded.
2. "When was this last run?" is already the artifact's COMMIT DATE —
immutable, verifiable, stated once. A generated_at inside the report would
be a second copy of one fact, drifting the moment a report is regenerated
and not committed. That is G-23's defect class, registered this same week.
3. It keeps the digest honestly deterministic. A timestamp would have to be
excluded from it — a value in the artifact that the artifact's own
integrity check ignores.
The contract also records what WOULD change this: a predicate whose claim is
about duration rather than sequence (H5's resource-cost leg is the live
candidate). B is already ruled, so that needs no new decision — only the
predicate that requires it.
The re-run: 5000 beats, reboot at 2500, under the POST-R-3 profile. All four
predicates pass. Committed as
evals/l10_always_on/results/report-5000-6c67e88d.json, digest f814fd97….
The most useful result was not the pass. Every DETERMINISTIC fact reproduced
exactly — vault bounded at 6, convergence at beat 1 with a 4999-beat tail,
clean reboot with derived learning intact — while the machine-variant float
moved from the prose's 1.389e-07 to a worst of 6.177e-08. That is the first
real evidence that excluding versor_condition from the digest was correct
rather than convenient: a digest covering it would have flipped on a machine
change and taught every reader to ignore it. The prose figure is superseded,
not contradicted — it was never a reproducible quantity.
tests/test_l10_soak_evidence.py, on the gate, guarding two DIFFERENT failure
modes that are easy to conflate:
REGRESSION — the digest is pinned, plus the horizon (a 24-beat report pinned
as though it were the 5000-beat one would pass everything else), the four
predicates by name (all_gates_pass alone is satisfied by an empty predicate
list), and a vacuity guard on the attested set.
STALENESS — the mode nothing guarded, and the one that had ALREADY FIRED.
R-9 ruled the cadence change-triggered, not clock-triggered: local-first
doctrine makes "nightly" a ruling rather than a cron line, and a nightly that
does not run is green for the wrong reason (H-9). So instead of "when did we
last run it?", the pin asks the question that matters — DOES THIS EVIDENCE
STILL DESCRIBE THE CODE THAT SHIPS? — by holding the SHA-256 of every source
the soak attests: chat/always_on.py, chat/always_on_daemon.py, and the three
harness modules.
That is not hypothetical. It had already fired silently for six weeks: the
2026-07-19 soak ran with accrue_realized_knowledge absent from
CONTINUOUS_LIFE_CONFIG_FLAGS, so from R-3 (PR-5, this same day) onward the
recorded result attested a process that no longer existed — and nothing in the
repository could notice. HAD THIS PIN EXISTED, PR-5 WOULD HAVE FAILED IT. That
is precisely the intended behaviour, and it is verified by sabotage: touching
always_on_daemon.py turns it red.
The failure message says RE-RUN THE SOAK — DO NOT EDIT THE HASHES, because
updating a hash to match changed source without re-running converts the pin
into a record asserting evidence that does not exist. That is the failure this
whole arc is about, committed inside its own remedy.
Three sabotages observed red: a touched daemon (the staleness mode), a
regressed verdict in the artifact, and a shorter horizon pinned as if it were
5000 beats.
H1-H4 promoted onto the gate, and the cheap option was deliberately refused.
Measured 20.95s for the four real-soak "holds" and 0.46s for the eight mutation
"bites" — 98% of the cost is the holds. Shipping only the bites, or parking the
file in a curated-but-unreachable suite, would have satisfied R-9's letter and
bought nothing; a suite nobody runs is the non-guarantee this arc spent PR-4
closing. Those four holds are the ONLY tests on any gate that run
chat/always_on's run_continuous end-to-end, and that loop just shipped a
six-week silent divergence. 21s per push to actually run the process that broke
is the most defensible time on this gate. Removed from full_only_baseline.txt
in the same edit; the baseline shrank 746 -> 745.
Closes G-5. Proof-of-life has moved from prose to committed, pinned,
change-guarded evidence.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Rank 3 on the docket. RuntimeConfig carries 32 boolean fields and nothing in
the repository stated that set, distinguished a capability flag from a posture
flag from a deployment knob, or recorded what evidence would flip any of them.
Two independent counts of the same set differed by eleven — the assessment said
"seventeen capability flags," N-7 measured 28 default-off plus 4 default-on —
and neither was wrong about what it looked at. They counted different things
because nothing declared what the set was.
R-3 (ruled A — incomplete flag set). accrue_realized_knowledge joins
CONTINUOUS_LIFE_CONFIG_FLAGS. Step D consolidates the realized facts Step B
writes, and the daemon forced the consumer without the producer: a continuous
life consolidating an empty set, which the mastery framework calls "garbage at
high speed." It shipped that way for six weeks.
The decisive evidence was in the code, not the documents. BOTH flags' comment
blocks in core/config.py already described the corrected profile —
accrue_realized_knowledge says "the production L10 process enables it alongside
persist_session_state", consolidate_determinations says "...alongside
accrue_realized_knowledge + persist_session_state". Three records (two
docstrings and the 07-25 verification doc) agreed with each other and disagreed
with four lines of code. Dormancy was a coherent ruling and would have cost
correcting three records that were RIGHT about the design; incompleteness cost
one line.
So the planned "N-6 docstring correction" deliverable DISSOLVED rather than
shipping: adding the flag made both docstrings true. Recorded explicitly,
because a PR that quietly ships less than it promised is the same divergence
class this arc exists to close.
R-4 (ruled A) — docs/specs/flag_register.md. All 32 booleans with class,
governing ADR or an honest "none", recorded rationale, and what evidence would
flip it. The classification is the load-bearing half and is stated as a
taxonomy rather than a label:
CAPABILITY — what CORE can do; engineering may flip it on named evidence
POSTURE — what CORE serves or refuses AS TRUE; the wrong=0 boundary,
ruling-only
DEPLOYMENT — cost, lifecycle, process shape; per-deployment
estimation_enabled and composed_surface are both "= False" and are not the same
kind of thing. Nothing said so before.
Profiles are declared as the UNIT of decision (R-4's mechanism): one-shot/eval
and continuous-life. The absence of a SERVING profile is stated deliberately —
the four POSTURE serving flags each ride their own license, and grouping them
before the ledger is sealed would license by association, the outcome R-8
exists to prevent.
tests/test_flag_register.py, on the gate. BIDIRECTIONAL, because either
direction alone would have passed through both failures it exists to prevent:
a flag added to config.py and not registered fails; a register row outliving
its flag fails. The counts the register's prose states (32 / 4 ON) are pinned,
since a stale count is how this register came to be needed. Four sabotages
observed red — new unregistered flag, deleted row, orphaned row, and a flag
silently flipped ON, the last caught by two independent pins.
THREE FINDINGS THE ITEM DID NOT ANTICIPATE, all from measuring rather than
reading.
1. Accumulated permissiveness, not just hesitancy. Of the four default-ON
flags, ONE has a governing ADR (deduction_serving_enabled / ADR-0256) and
TWO have no recorded reason of any kind — allow_cross_language_recall and
use_salience carry no comment block, no ADR, no criterion. In an
architecture built on earned licenses, two permanently-on capability flags
with no recorded decision is G-8's question inverted. Registered and
deliberately NOT fixed: writing a rationale after the fact would invent a
decision nobody made, which is worse than recording the absence.
2. A hollow gate inside the governance pin itself.
test_default_on_flag_is_not_governed_by_a_proposed_adr is parameterized over
(default-ON flag x cited ADR), so the three uncited ON flags contribute ZERO
cases — it covered exactly one flag. Its non-vacuity guard asserted only
that SOME flag cites an ADR, which default-off flags satisfy in abundance.
One reformatted comment away from zero coverage with everything green: the
failure state indistinguishable from the success state, inside the module
written to prevent exactly that. Guard tightened to require a non-empty
parametrization, and observed red by stripping ADR-0256's citation.
3. The daemon's own tests ran on no gate. R-3 changes what the always-on
process does every beat, and tests/test_l10_always_on_daemon.py was in no
curated suite — so this behaviour change would have shipped guarded solely
by tests no gate invokes, which is the hollow-gate pattern landing on the
exact file covering the change. Promoted to smoke WITH the change it guards.
Measured +9.4s, +4.3% of smoke — paid deliberately, not estimated. Removed
from full_only_baseline.txt in the same edit (both-directions ratchet; the
baseline shrank 747 -> 746).
Red before green throughout: the assertion cfg.accrue_realized_knowledge is
True was added to the daemon test and OBSERVED FAILING before the flag existed.
H-8(c) closed by AMENDING the correction note rather than deleting it, and the
amendment is the more useful artifact. PR-1's note said the 07-25 document's
claim was false; R-3 then made that claim TRUE, so the correction note itself
went stale — in the direction nobody guards, a correction made wrong by its own
subject being fixed. Both notes now carry the commit window they apply to. A
dated correction survives its subject changing; an undated "this is false" does
not.
Also delivered, from the 2026-07-28 external-assessment triage: §5, the
declared-table index. Eight single-source-of-truth tables and the pin that makes
each true. A reader's aid, explicitly NOT a central contracts.toml — that would
add a fifth copy needing agreement with four generators and put four different
authorities in one merge surface.
Handed forward to PR-11: the 5000-beat soak recorded in
evals/l10_always_on/contract.md (2026-07-19) predates this change and therefore
describes a configuration that NO LONGER SHIPS. PR-11's re-run is the first soak
of the corrected profile, and Step B feeding Step D is exactly what a long
horizon stresses.
Closes G-6, G-8, H-6, H-8(c), H-8(d) — and H-8 in full. Wave 4's F-6 half-gate
is lifted; the Track C half is untouched.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Rank 1 on the adopted docket, and the cheapest item on it: two ratified ADRs
whose records contradicted the code, plus the ADR-0252 obligation that Track
A's ratified NO-GO created. All four land in two files because they are two
files under one authority — splitting them would have bought two reviews of
the same document.
ADR-0146 (R-12a, option A) — the daemon it rejected, shipped.
The Shape-A rejection stands AS REASONING: a background daemon does require
process-supervision infrastructure that Shape B does not. That infrastructure
was subsequently built. chat/always_on_daemon.py implements the three items
"What is NOT in Scope" excludes, each mapped to where:
- cross-process file locking -> advisory fcntl.flock(LOCK_EX|LOCK_NB) at :82,
acquired BEFORE any signal or runtime setup so a second life fails fast
- signal handling -> SIGINT/SIGTERM graceful stop at :145
- daemon synchronization -> the load-time strict-identity guard; the flag is
forced at :48 and a different-identity checkpoint is refused, so a restart
is the same life or it is nothing
Shape B remains the persistence model and the CLI default; the daemon is an
additional shape layered on it, owned by this ADR. A new ADR for a shape that
shipped six weeks earlier adds a document without adding a decision.
The stale bullet gets an INLINE pointer, not only a trailing addendum. H-8's
mechanism is that an authoritative line converts "I should check" into "I
already checked" — the reader who stops at the excluded-scope list is exactly
the reader this gap is about, and they never scroll to the end.
ADR-0252 (R-12b, option A) — the "34 organs" basis footnote.
Footnote at the first use, marking the count as underived: the reproducible
figure is 18 resolve_promotable_* organs, ALL in generate/derivation/ and none
anywhere else, with exactly 32 modules in the package as the likely origin of
"34". The diagnosis, supersession plan and §4 conformance bar are unaffected —
none depends on the count. Option B (replace every "34" with "18") rejected: it
rewrites a ratified document's prose where a note at the point of use is all
that is owed.
ADR-0252 §5 — the verdict banner. NO-GO, ratified. Criterion pre-registered at
299c92be BEFORE the run, run at 797ebad5, artifact digest b3d9d275…, with the
mechanism stated in one measured sentence: the similarity quotient that
delivers attribute-invariance is the same quotient that annihilates structural
contrast. Scope held narrow on purpose — this refutes H1 for point-position
role encodings aligned under similarity, not every Cl(4,1) representation, and
not the symbolic lane. Also records N-8 (two prior GO verdicts on unmerged
branches, both unsound) and G-21.
ADR-0252 §6 — the amendment the NO-GO forces, and the load-bearing half.
§6 retired the surface organs "until the structure-mapper proves out" and named
no replacement for the NO-GO branch. Left alone that is a live instruction that
can never fire — an authoritative-looking dead instrument, the H-9 class
exactly. Amended: the organs are RETAINED, and that is the operative state
rather than a pause in a transition; the retirement condition is re-stated as a
demonstrated replacement passing the §5 criterion, which does not exist. The
"Build (gated on §5 GO)" bullet is marked NOT AUTHORIZED with the reason.
Retiring a working organ on the strength of a hypothesis ABOUT its replacement
is backwards; the hypothesis has now been tested.
Verification — every factual claim re-derived at ca5e614e rather than copied
from the packet: 18 organs, 0 outside generate/derivation/, exactly 32 modules
(tightening the draft's "~32"); lock, signals and identity guard each cited by
line; report digest read back from the artifact file.
Three things caught before landing, recorded because this commit is about
records being wrong.
1. The date. The addendum drafted "landed 2026-06-14". This container's clone
was SHALLOW — 168 commits, roots grafted at 2026-07-19 — so git log
reported 2026-07-20, which is the shallow boundary and not a fact. Neither
figure was verifiable until the clone was deepened to 2340 commits, which
confirmed the packet: 18e25580, 2026-06-14, the same commit introducing
both the lock and the signal handling. The addendum now cites the SHA. A
date that cannot be re-derived is testimony.
2. A wrong section reference — the amendment cited §9 for a sentence in §8.
3. A misquote — "the existing reader" for "the existing 34-organ reader".
Writing a misquote into a ratified document while amending that document for
accuracy would have been this ruling's own failure mode committed inside its
own fix. Both were found by re-reading the amendment against the source it
quotes, which is standing philosophy 7.
Records closed: H-8(a), H-8(b), G-15. Track A's outstanding obligation
discharged. H-8(c)/(d) remain open on R-3 and land with PR-5.
No runtime behavior. No test changes. Two ADR files and four register/plan
files.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
core/capability/domains.py's DOMAIN_PACKS declares domain -> pack membership
for the capability ledger. Each packs/data/<pack>/manifest.json independently
declares domain_id. Two statements of one fact, and only one direction was
ever checked: domain_contract_predicates P3 validates domain_id -> a known
ledger domain. Nothing validated DOMAIN_PACKS -> the manifest agrees.
P3 passes VACUOUSLY on a manifest with no domain_id at all — the same shape as
its own test_pack_without_contract_reports_absent. So a pack that DOMAIN_PACKS
places in a domain, but whose manifest never says so, is invisible to every
check in the repository.
Measured: 7 of 9 bound, 0 contradictory, 2 absent — en_core_cognition_v1 and
en_core_meta_v1, both placed in philosophy_theology, which is one of the four
bands queued to earn a SERVE license behind R-8. The domain whose license is
next to be earned is the one whose binding exists in only one place.
Why nothing caught it: tests/test_domain_contract_predicates.py exercises
P1-P9 entirely against synthetic tmp_path manifests, never against the real
nine, and is itself in no curated suite. A predicate proven correct on
fabricated input and never run on the real input is this arc's defect class
stated precisely — its success state and "it never ran here" are
indistinguishable.
tests/test_domain_pack_binding.py, registered in smoke (on the gate):
- every pack DOMAIN_PACKS names has a manifest on disk
- a manifest that CONTRADICTS the table always fails
- a manifest with NO domain_id fails unless the pack is in UNBOUND_BASELINE
- the reverse direction: a pack may not self-assign a ledger domain the
table does not grant (clean across all 30 packs today, pinned because a
new manifest copying domain_id from a sibling is the obvious mistake and
would silently widen a licensed domain)
- both directions on the baseline: a pack that gains domain_id must leave it
- the parse is guarded against vacuity (N-1 has produced three false
conclusions in this repo)
- the count is pinned so the gap moving is a reviewed decision
Four sabotages, each observed red, each caught by its intended pin.
PINNED, NOT FIXED, deliberately. Whether en_core_cognition_v1 is a
philosophy_theology domain pack or merely grouped under that domain for ledger
accounting decides what the license actually covers. That is
ratification-adjacent content, not engineering, and it is routed to R-8 /
PR-14 where that band's license is decided. Sealing that ledger without
answering it licenses a domain whose membership is asserted in one place and
denied by silence in the other.
How it was found, recorded because the method is the point. An external
assessment proposed collapsing the pack/domain machinery, stating its method
up front: "I have enough from the directory and CLAIMS.md to reason
seriously." That is the method AGENTS.md protocol rule 1 forbids, and five of
its six proposals are falsified by opening the file — recorded in full in
31-hindrance-audit.md, in short:
- conftest.py "almost certainly" holds suite-membership logic: it holds two
autouse fixtures and two data registries, and its collection hook stamps
markers and never skips. Zero membership logic.
- cli_test.py "almost certainly a nested conditional or a big dict": a flat
dict, 17 keys, zero conditionals in the module body.
- field/algebra/morphology as a peer triad needing a single serve(): 72, 15
and 3 importers respectively, morphology is 97 lines, and no def serve(
exists anywhere in core/, chat/ or generate/.
- the grade space is richer than the serve layer consumes: all 15
EpistemicState members have non-test consumers.
- constraints: blocks in 314 ADRs: that is 314 new places that must be kept
true, inside ratified immutable records, in the repository whose dominant
defect class is records drifting from reality. The existing design puts
the machine-readable half in the checker that reads the ADR, which is
better on purpose.
- a root contracts.toml as "the highest-leverage single change": the
principle is already ADR-0263 Rule 5 and is implemented in the very file
the assessment cites. Centralizing four tables with four authorities adds
a fifth copy that must agree with four generators.
The seam it pointed at was real even though every specific was wrong. Bad
method, useful direction; both halves recorded, and the triage rule earns a
second demonstration — every rejected proposal failed on a file its author did
not open.
AGENTS.md — the standing philosophy revised to eleven, per Shay's review.
- "be quick to be humble" -> "investigate yourself first". Speed is a
consequence of reflex, not a criterion; writing it in makes velocity sound
like the point. Examples replaced with three that are verifiable in-repo:
the H-13 refcount claim, and the two rulings that were about to be
recorded stronger than they were made.
- the defect-class principle gains its prescriptive half as its own
sentence, and gains this commit's instance: a predicate that passes
vacuously on the input it was written for.
- NEW #2: don't build a ratchet to guard a ratchet. Placed adjacent to
"less is more" and named as its corollary so it does not read as
duplication. The arc paid for it — a membership ratchet needed a
reachability ratchet to mean anything, and what actually shrank the gap
was deleting four suites nobody called.
- NEW #10: a closed issue is not evidence the problem is gone. G-7 was
CLOSED and REOPENED-PARTLY inside one arc; H-13 was closed with a fix that
did not work.
Two review points were already satisfied in the file and are recorded as such
rather than re-applied: the defect-class principle already stood alone on its
first line, and already sat at position 3, not 8. The review was reading a
chat summary, not AGENTS.md — which is, exactly, reviewing the record instead
of the reality.
PR-5 additionally owes one page: an index of which declared tables exist and
what pin makes each true. That is the defensible residue of the contracts.toml
proposal, and explicitly not a fifth copy of four tables.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Shay adopted twelve rulings, struck two, and ratified the ADR-0252 §5 NO-GO.
For several hours afterwards the packet still said PENDING fourteen times,
while PR-3 and PR-3b had already executed against R-7. That is precisely the
failure this arc is about — a record contradicting reality at a load-bearing
point — committed inside the arc whose subject is that failure, and it was
caught by a question rather than by any instrument. Fixing it is the first
item here; the rest of the commit exists so the same gap cannot reopen by
context loss.
50-rulings.md
All fourteen statuses replaced. Twelve carry RULED <choice> with the date
and the deciding authority; R-10 (discharged before the arc) and R-14
(dissolved by N-9) are struck. An adoption block at the head records the
choices, the strikes, the §5 ratification, and the corrected execution
order, plus which rulings have already executed.
50-execution-plan.md
New §2.1 — the adopted docket. The twelve choices, and the execution order
as a ranked table stating why each item sits where it does. The packet
ordered by register number, which interleaves cheap record repairs with
expensive capability work and puts the two zero-risk ADR amendments last;
the corrected order is by what each item unblocks, cheapest first in tier.
Three things that lived only in conversation are now written down:
- the claims register is HELD, deliberately, with its entry criterion —
G-22 showed CLAIMS.md can publish a superseded digest for two days, and
an index over untrustworthy indexed things is the H-9 pattern;
- gate width is promoted from a parenthetical to PR-4b, sized, with
`teaching` recommended alone and the reason being evidence (G-22) rather
than taste — a parenthetical is how fifteen suites became unreachable
without anyone choosing it;
- Track A's ratification creates an obligation the verdict document cannot
discharge: ADR-0252 §6 retires the surface organs "until the
structure-mapper proves out," names no replacement for the NO-GO branch,
and is an instruction rather than a research note. A live instruction
that can never fire is a dead instrument.
Two places where I was about to state a ruling more strongly than it was
made, corrected rather than smoothed over:
- R-11 B is "measure, then re-ask." It authorizes no gate. It owes a
measurement AND a second ruling, and a ruled-but-inert item looks exactly
like a settled one at a glance.
- R-8 C was recommended "with a floor from A applied to the entailed
capability only" and N was never named. PR-14 must carry a proposed N with
its consequence, or land entailed as explicitly unlicensable-pending-N.
Shipping without a floor would license entailment on nine cases through an
omission — the exact outcome C exists to prevent.
Also: §0 corrected from seven corrections to nine (N-8, N-9 were added
during execution and the heading never moved); definition-of-done item 5
struck, because N-9 killed both its halves — smoke.yml parity is not a goal
under AGENTS.md:280, and "no test file hides in full" is unreachable at 747
files without ceremony that changes nothing about what executes.
30-gap-register.md
G-7's open half is 15, not 19: PR-3b took it there by deletion, which costs
no gate time. Points at PR-4b.
AGENTS.md
Nine standing philosophies under the reasoning protocol, per CLAUDE.md's
rule that governance and memory live here. They are the generalization of
what this arc actually cost: less is more; master the small details and the
large ones do not form; a mechanism whose failure state is indistinguishable
from its success state is not a guarantee; a record that contradicts nothing
beats one that impresses; consecutive unclear problems mean the frame is
wrong, not the fix; be quick to investigate yourself, and record the error
next to the fix rather than overwriting it; volume of work is not a licence
for lower diligence; fix what you find or name why not; let the tree settle,
then run the gate once.
Documentation only — no runtime behavior, no test, no pin changes.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Wave 1 — delete — run before more Wave 2 enforcement, correcting the method
inversion this arc had been committing: the plan sequences scrub -> DELETE ->
enforce, and Wave 1 was ruling-blocked while Wave 2 was not, so enforcement
landed first on structures deletion might have removed.
PR-3 (R-7 ruled A)
docs/gaps.md and docs/audit/substrate-liveness-ratchet.md carry historical
banners that name H-9's own mechanism as the reason they exist: an instrument
that looks authoritative converts "I should check" into "I already checked".
The ratchet's seven OPEN items — W-003, W-005, W-007, W-008, W-009, W-017,
W-018, every one L10-chained — migrated into G-5 WITH their dependency
chains, because that reasoning is the ratchet's lasting contribution and its
status column is not.
One deliverable could not be executed as written and is recorded rather than
claimed: L12 has NO in-repo generator to drop it from. The system map is
local and gitignored (D5 — a regeneratable index carrying no authority), so
the phantom existed only there and in one taxonomy row. That row now records
the ruling instead of the flag.
PR-3b — new, and it corrected its own author
Added when PR-4's two ratchets made the question unavoidable: what exactly
are they policing? Measured: 21 curated suites, 9 holding exactly ONE file,
12 holding <=4, and 2 reachable from any gate.
I proposed cutting ~11. The evidence supported FOUR. cognition, teaching,
packs and algebra are named in AGENTS.md's own pre-merge-gate instruction;
fast, pulse and proof in the CLI's help text; phase5, phase6, adr-0024, math
and formation in READMEs and ADRs. Deleting those breaks documented commands
— worse than the sprawl. Taking the 4 the evidence supports, rather than the
11 I had already said out loud, is the whole point.
Deleted: refusal, margin, rotor, inner-loop. Zero --suite references anywhere
in code, docs, ADRs, workflows or CLI help. Per-phase ADR-0024 aliases
offered so reviewers could run a phase independently; nothing ever did. All
seven of their files remain covered by adr-0024 — verified BEFORE the cut —
so nothing was orphaned and no coverage moved.
21 -> 17 suites. Pin 3's gate-unreachable gap 19 -> 15 BY DELETION, which
costs no gate time at all, versus promotion which costs it every run. An
alias nobody calls is not a curation decision; it is one more name that has
to be kept true, and two ratchets were policing these.
Registers: H-9 CLOSED, PR-3 and PR-3b marked LANDED, G-5 carries the seven
migrated items, the taxonomy's L12 row records the ruling.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
PR-4's membership ratchet closed "no new orphans". Within the hour this arc's
first full-tree run found tests/test_ratification_ceremony.py RED, in the
curated suite `teaching`, invoked by NO gate tier. Membership was satisfied and
bought nothing — which is exactly the objection PR-4 itself raised against
G-7's original formulation: a curated suite nobody runs is the same
non-guarantee as `full` wearing a different name. The fix had the same blind
spot as the thing it fixed.
Measured: 21 curated suites, 2 gate-reachable (smoke, deductive). The gap is
NINETEEN, not the handful first estimated.
Shipped as baseline + ratchet, sized to the measurement rather than the rule:
19 `post-merge-only` justifications written in one sitting is the ceremony
failure already rejected for the 749-file membership baseline. The pin freezes
the gap, fails on growth, enforces both directions (a promoted suite must leave
the baseline), pins the count, and — critically — VERIFIES ITS DECLARED GATE
SET AGAINST THE SHELL (pre-push / local-ci.sh), so it cannot go vacuous the way
a scan that silently matches nothing does (N-1's class, three false conclusions
in this repo so far).
Two sabotages observed red: a new curated suite with no gate caller, and a
GATE_SUITES value drifted from the scripts it claims to describe. The second
correctly cascaded into two failures, which is the anti-vacuity guard working.
What pin 3 deliberately does NOT do: decide which of the 19 belong on the gate.
That costs gate time and is a decision, not an engineering task. Recorded as the
open half of G-7. `teaching` is the concrete first candidate — it just shipped a
red ratification pin that no gate ran.
Process note: a gate run was invalidated by editing the tree mid-run for the
third time today. Stopped with TaskStop rather than pkill this time, which
reaped cleanly instead of leaving the orphaned wait-loop that ran for 2h16m
earlier. All edits now land before any run starts.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Ruled: fix, don't document. The argument that settled it — a red test on a
serving-path contract left with an explanatory note IS the H-8 failure mode, a
document acknowledging a gap instead of closing it, which becomes the next
assessor's finding.
Two distinct failures, and the second is the serious one.
1. THE CEREMONY PINS WERE STALE AGAINST CORRECT CODE (H-8e's class).
ChainRecord.polarity defaults to AFFIRMATIVE and is DELIBERATELY OMITTED
from affirmative rows (ADR-0264 R1; teaching/ratification.py:119). Both pins
built ChainRecord(**{k: row[k] for k in __slots__}), so they KeyError'd on
every correctly-serialized row. Now built from the row's own keys, letting
absent fields take their dataclass defaults — which RESTORES the guarantee
rather than weakening it. Consequence of the staleness: ratification-corpus
byte compatibility and unreviewed-status refusal were UNVERIFIED since
ADR-0264 R1 landed.
2. CLAIMS.md PUBLISHED A SUPERSEDED EVIDENCE DIGEST FOR A LICENSED CAPABILITY.
f9e9cc0c (2026-07-26, "the deduction lane hashes the prose it serves")
strengthened the lane, rewrote evals/deduction_serve/report.json, and moved
the authoritative pin in scripts/verify_lane_shas.py:60 to c855d55c —
recording the superseded 0b461a5a on line 56. It did NOT regenerate
CLAIMS.md. So for two days the published claim for deduction_serve_v1
pointed at evidence that no longer existed. The claim TEXT never changed;
only the pointer was wrong — which is exactly the failure a digest exists to
prevent. Regenerated; the diff is one line, the digest.
The drift had reached a THIRD artifact: this assessment's own
M3-comprehension-reasoning.md card cited the same stale 0b461a5a as its
would-fail-if-absent evidence. Corrected.
THE MECHANISM IS G-7's, AGAIN. One commit updated the lane, its verifier and
its lane test, and missed one downstream artifact — and the pin that guards
that artifact (tests/test_claims_md_is_current.py) EXISTED AND RAN ON NO GATE.
Nothing surfaced it until this arc ran the full tree once. Both it and
test_ratification_ceremony.py are now registered in TEST_SUITES["smoke"].
The membership ratchet caught the author a second time: promoting
test_claims_md_is_current.py onto the gate without removing its baseline line
failed test_baseline_has_no_stale_entries immediately. Baseline 748 -> 747.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
The arc's first full-tree run — 13,403 passed, 3 FAILED — and the three
failures reproduce identically at 797ebad5, the arc's starting commit. NOT
arc-caused. main has been red for an unknown period and nothing surfaced it.
tests/test_ratification_ceremony.py::test_row_is_byte_compatible_with_the_committed_corpus
tests/test_ratification_ceremony.py::test_unreviewed_status_is_refused
tests/test_claims_md_is_current.py::test_claims_md_matches_generator_output
DIAGNOSIS — the code is right and the pins are stale, same class as H-8e.
ChainRecord.__slots__ gained "polarity" (ADR-0264 R1), and the serializer
DELIBERATELY OMITS it for affirmative rows (teaching/ratification.py:119:
"polarity is OMITTED when affirmative"). A test that rebuilds
ChainRecord(**{k: row[k] for k in __slots__}) therefore KeyErrors on a
CORRECTLY serialized row. Consequence: the ratification-corpus byte-
compatibility guarantee and the unreviewed-status refusal have been UNVERIFIED
since ADR-0264 R1 landed.
THE STRUCTURAL FINDING, which is bigger than the three tests. test_ratification
_ceremony.py IS in a curated suite — `teaching` — and no gate invokes it. The
pre-push gate runs smoke + the warmed_session lane + deductive. So a curated,
red, governance-path pin sat unrun.
That is this arc's own hollow-gate objection turned back on its own fix. PR-4's
ratchet closes MEMBERSHIP. It does not close EXECUTION, and I closed G-7 as
though it did. G-7 is reopened-partly and PR-4 now owes a third pin:
PIN 3 — every curated suite must be reachable from a gate tier
(scripts/hooks/pre-push / local-ci.sh), or be declared post-merge-only with a
stated reason. Currently unreachable: teaching, packs, algebra, sensorium,
cognition, runtime and the rest — they have members and no caller.
Membership was never the guarantee. Execution is.
Recorded as G-22 with the diagnosis; repair is test-side and mechanical, but
these guard governance-path behavior and a published artifact (CLAIMS.md), so
the disposition is Shay's rather than mine.
Docs only.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
§9.8 of the plan — "every card carries a verified_at no older than the last arc
that touched its layer" — is a maintenance contract this directory holds itself
to, and today's arc moved MV, M3, M4, M2, M6 and three component cards. A card
whose stamp falls behind a load-bearing arc is testimony, not evidence, which is
the failure mode the directory exists to replace.
Each card gains a Re-verified stamp at 39331dbc plus an arc note saying what
actually changed under it, not merely that something did:
MV PR-4 + PR-6 landed; G-7/G-9 closed. The layer's finding is sharper than
the card stated: the recurring defect is not ABSENT pins but PINS THAT
EXIST AND DO NOT RUN — four found this arc, each indistinguishable from
coverage in any document.
M3 §5 run to NO-GO, settling the successor-path question. New measurement
G-21: the reader decides 5 of 500 holdout_dev/v1 cases. H-8e corrected.
M4 H-13 fixed — the served-marker honesty contract held where it leaked.
M2 PR-7's table: two boundaries already met, THREE CLOSED as category
differences, one real delta (no audit record per rejection).
M6 PR-9 — the accrual backstop counts and names what it absorbs.
realize-phase / always-on-process / surface-selection — the component-level
versions of the same three.
Plan §Status replaced with the actual board: what landed, what each remaining
item is blocked on, and the six findings this arc added that the assessment did
not have (N-8, N-9, G-21, H-13, H-14, H-8e). README carries the same summary at
the directory's entry point.
The status block says plainly what has NOT moved: comprehension breadth and
proof-of-life. The reader is still 19 constructions wide and decides 1.0% of the
held-out corpus; the 5000-beat soak is still prose with no pinned digest.
Everything landed this arc is enforcement and evidence machinery. Capability
moves in PR-11/PR-12/PR-14/Track B, all four ruling-gated.
Docs only.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
formation/ declares its trust boundary explicitly (forge.py's five ordered
rules + hashing.py's content-addressing, no-floats, no-pickle, self-sealing).
ingest/gate.py — the PRODUCTION boundary, the only point where raw data enters
the versor manifold, facing untrusted user text every turn — had no comparable
table. H-7: asymmetric rigor invites the assumption that the un-tabled boundary
is the less important one, when it is the opposite.
Written per H-7's authority: documentation first, evidence decides whether code
follows. Every row measured against the code at edf6c2a4, not assumed — and the
measurement reframes the entry.
ALREADY MET (2) no floats in hashed payloads — _stable_digest hashes
salt + name, both UTF-8 strings, no float reaches a digest
no pickle — absent from all of ingest/
CATEGORY content-addressed artifacts, self-sealing, source
DIFFERENCE (3) allow-list. All three presuppose a PERSISTED ARTIFACT WITH
CITATIONS. formation validates artifacts it will store; the
gate transforms text it will not. Recording these as owed
obligations would have manufactured three hardening tasks
that cannot be done, and sent a future reader to do them.
Closing them is the load-bearing half of this table.
REAL DELTA (1) no audit record per rejection. The gate refuses with typed
exceptions (KeyError :235/:286, RuntimeError :318/:424) and
records nothing about what it refused;
vocab.record_unknown_token :331 records OOV observation,
which is not a rejection record.
On the delta, stated so it is not over-read: it is NOT a containment hole —
refusal is fail-closed and nothing unvalidated crosses into the manifold. It
costs forensics. It is the same failure class as H-11, which PR-9 closed on the
turn spine hours ago: a guard that works correctly and leaves no trace of having
worked. And it has a named downstream consumer — a widening program (G-3) must
be able to ask "what shapes is this boundary refusing, and how often", which
today it cannot.
Hardening for B6 is a separate PR per this entry's own authority, sized against
whether anything consumes gate-rejection telemetry today.
AGENTS.md §Security-and-trust-boundaries defaults also checked row by row for
completeness: four are n/a to a component that executes nothing, touches no
filesystem, and mutates nothing; the display/log default is met.
Docs only. Registers: H-7 CLOSED, PR-7 marked LANDED.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
PR-6 promoted tests/test_safety_pack.py onto the gate and left it listed in
tests/full_only_baseline.txt. test_baseline_has_no_stale_entries failed the very
next smoke run, naming the file and the direction.
That is the both-directions half of the ratchet doing exactly the work it was
written for, roughly an hour after it landed, against the person who wrote it.
A baseline that silently keeps entries it no longer needs stops being a
measurement and becomes a story about one — so the failure is the feature.
Recorded in the pin's own docstring and in G-7, because a mechanism catching its
author is better evidence than a sabotage test: the sabotage was contrived, this
was not.
Baseline: 749 -> 748. The ratchet only turns down.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
G-9's charge: "law-enforced-by-review is weaker than law-enforced-by-test." Its
stated method — verification pass first, then mechanical PRs — is what made this
worth doing, because the three prohibitions turned out to be in three different
states and only one of them was the state the register assumed.
(a) EXACT RECALL — the real gap, and worse than recorded. "Would a cosine
ranker actually fail a test?" answers NO across nearly the whole surface. The
only import ban in the repository covers exactly one file
(core/physics/wave_manifold.py, via test_third_door_cohesion.py — itself in no
curated suite). Meanwhile generate/realize/recall.py, which IS the recall path,
states the prohibition in its module docstring — "an exact, deterministic
equality scan (no cosine / HNSW / ANN)" — and nothing enforced it. Law asserted
where it executes, enforced nowhere: H-8's failure mode applied to a
prohibition rather than a record.
Now scanned across vault/, generate/realize/, generate/meaning_graph/, field/,
recognition/ for banned ANN libraries AND for hand-rolled cosine* definitions.
The second half matters: an import ban only stops the convenient version, and
ten lines of numpy is the realistic way a well-meaning contributor violates
this. A third pin ties the docstring claim to the scan, so the sentence and its
enforcement cannot drift apart.
(b) GOVERNANCE BYPASS — now exact. INV-07 proves governance works WHEN CALLED;
nothing proved it could not be skipped. Mechanism: no function in
chat/runtime.py may construct a TurnVerdicts without invoking
safety_check.check, and every such call must pass self.safety_pack — because a
bypass need not skip the call, it can substitute the pack. Measured at close:
two verdict-constructing functions, both governed.
(c) SAFETY-PACK NON-SWAPPABILITY — already well pinned; the gap was REACH, not
coverage. tests/test_safety_pack.py pins unratified-pack-refused-in-production,
missing-companion-report-refused, seal-failure-refused, path-traversal-rejected,
missing-pack-fails-closed — and ran in NO curated suite, so a fail-closed safety
contract was verified only after merge. Promoted onto the gate; the new pin
guards that placement rather than duplicating the substance.
FOUR SABOTAGES, EACH OBSERVED RED, each caught by its own pin: a banned library
imported on a recall path, a hand-rolled cosine ranker, a serving path
assembling verdicts without governing, the safety pack demoted off the gate.
Both new files registered in TEST_SUITES["smoke"] on creation (#136). Source
scans only — no runtime behavior, no flag, no serving path touched.
Registers: G-9 CLOSED with the per-item findings; PR-6 marked LANDED.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
N-9 is the finding, and it PREVENTS work rather than re-scoping it.
WHAT THE PLAN GOT WRONG. N-3 concluded the 23-vs-13 smoke delta was unintended
drift, on two in-repo comments asserting parity, and called that "the strongest
evidence available that the drift was never intended". H-12 repeated it; R-14
built three options on it recommending "raise CI, +46s measured"; PR-4
specified a bidirectional parity pin as deliverable one.
A third artifact answers all of it, and none of them read it.
tests/test_cli_test_suites.py::test_cli_smoke_suite_covers_ci_smoke_gate has
pinned smoke.yml SUBSET-OF TEST_SUITES["smoke"] since before this assessment —
which is why every measurement finds CI-only == 0; a pin, not luck — beneath a
comment headed "DELIBERATELY ONE-DIRECTIONAL — do not 'complete' this by also
asserting local_paths <= ci_paths". Its history is 50fa287d (2026-07-25):
"revert(tests): drop the local-to-CI smoke parity assertion — CI is not a
gate." An earlier agent read the same comment, drew the same inference, made
the assertion symmetric, and reverted it hours later: "That was wrong, and
AGENTS.md says so in a line I had already read."
AGENTS.md:280 — GitHub is "a mirror only; its Actions are billing-locked and
produce dead signals — never chase them." §277 — the workflows are "secondary
observability only." So the local suite is the SOURCE and is deliberately
broader; the delta is the design.
- PR-4 pin 1: WITHDRAWN. It exists, in the direction that protects something.
- R-14: premise corrected in the packet. Option A spends 46s on a workflow
that does not gate; option B would delete real protection for a fiction;
option C's live half is a two-line comment fix, done here.
- N-3's exposure claim was too GENEROUS to CI: for a push that skipped the
local gate there is no automatic check at all. Larger than stated, and not
closable by editing smoke.yml.
WHAT LANDED. Measured: 749 of 877 test files are in no curated suite. The
plan's mechanism ("assign every orphan, or 749 exclusion reasons") is hollow at
that scale — glob topic-suites ("adr": tests/test_adr_*.py absorbs 172) satisfy
the assertion while changing nothing about what executes, which is precisely the
Third-Door objection the plan levels at G-7's own first formulation. All four
demonstrated incidents (#113, #136, negation-survives-articulation,
speculative-subject-lifecycle) were caused by a NEWLY LANDED orphan; none by a
legacy one.
So: a ratchet. tests/test_suite_membership.py + tests/full_only_baseline.txt.
New orphans fail. The baseline is enforced in BOTH directions — a promoted or
deleted file must leave it, so the list can only shrink — and its count is
pinned so bulk movement is a reviewed decision, in the discipline
test_volume_honesty.py uses.
Also: the gate-guarding pin now runs on the gate. test_cli_test_suites.py lived
in `fast`; the pre-push gate runs smoke + deductive. The pin guarding the gate
did not run on the gate.
Three sabotages, each OBSERVED RED: a new unregistered test file, a stale
baseline entry, the parity pin demoted out of smoke.
Note on the delta: now 12 (local 27, CI 15) because this session added three
pins locally. Under the corrected reading that is the design, not widening
drift — the invariant is CI-only == 0, and that is pinned. The registers now
say so; "ten files" was a measurement, never a target.
Process note: `git checkout --` during sabotage cleanup destroyed uncommitted
edits to core/cli_test.py, which were redone. Backups, not checkout, on files
with unstaged work.
Registers: G-7 CLOSED, H-12 amended, R-14 dissolved in the packet, PR-4
re-specified, N-9 added to plan §0.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Wave 2's PR-9, whose authority is "mechanical PR" and which is gated on no
ruling. Widened by the 2026-07-28 triage from the accrual guard alone to all
three broad `except` blocks on the turn spine.
Each guard is DELIBERATE and stays: accrual is additive and must never crash a
turn; a pack/catalog failure must not take down logos authority; a probe
failure must not touch the served surface. Narrowing them to ImportError/
OSError — as an external assessment proposed — would convert a malformed
decision object into a turn-spine crash, trading silent failure for
served-surface failure. What was wrong is not the tolerance, it is that each
wrote a value INDISTINGUISHABLE FROM ITS HEALTHY CASE:
- a raising read->realize->determine chain left the same None accrual a quiet
turn leaves;
- a logos decision that raised looked exactly like a turn where logos was
never consulted (both: logos_decision_kind == "");
- a crashed geometric probe recorded the same empty neighborhood as a probe
that ran and found nothing — in the block whose entire job is producing the
data that prices the anti-unification roadmap.
Silence, in the one layer whose constitution is "failures are typed, never
silent" (INV-34).
chat/runtime.py _accrual_swallows / _last_accrual_error, read via
accrual_swallow_telemetry(). _last_turn_accrual is
still None on failure, so _maybe_surface_determination
returns byte-identically — ADDITIVE BY CONSTRUCTION,
and pinned as such.
cognition/result.py logos_error on CognitiveTurnResult.
cognition/pipeline.py logos_error recorded; probe_error,
neighbor_scan_errors, graph_anti_unify_error on the
OOV telemetry.
tests/test_accrual_swallow_telemetry.py — 7 pins. The two behavioral ones were
OBSERVED RED against a build carrying the fields with the recording stripped,
so they discriminate rather than merely pass. Two test bugs of mine were fixed
along the way (wrong logos module path; CognitiveTurnResult has required
fields) — the tests were wrong, not the code.
Registered in BOTH TEST_SUITES["smoke"] and smoke.yml on creation, per #136.
Local/CI delta verified unchanged at exactly ten (local 25, CI 15, CI-only 0),
so this settles nothing R-14 has to rule.
Cleanup as found: three dead imports removed from chat/runtime.py
(_COGNITION_PACK_ID, _TEACHING_CORPUS_ID, engine_identity_for_config), each
confirmed to have no re-export dependence first — workbench/readers.py imports
engine_identity_for_config straight from core.engine_identity. ruff 4 -> 1 on
these paths; the survivor is a pre-existing structural E402.
Registers: H-11 CLOSED; PR-9 marked LANDED in the plan.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
71542e61 fixed only half of H-13 and asserted a premise that is false. This
replaces it, and records why, because the failure mode is the one this arc
exists to catch.
WHAT WAS WRONG. The refcount rested on "seeding and eviction walk the same
source list, so the counts balance." They do not. The call site's arms are
MUTUALLY EXCLUSIVE (pipeline.py §10b): a proposal is seeded when SPECULATIVE or
released when COHERENT, never both. So a release never balances a claim the
same proposal made, and a COHERENT proposal ALWAYS releases claims it never
held. Refcounting therefore repaired only the sub-case where both proposals had
been taught speculatively. The ordinary case — an unreviewed `wisdom`, then a
separate correction about `practical wisdom` that passes review on its own turn
— still stripped the marker. Measured against the landed refcount:
`sibling marked? False`.
The first attempt's tests passed because their helpers split tokens themselves,
mirroring a call pattern production no longer used — tests written to the fix
instead of to the defect. That is the same "asserted past the evidence" failure
diagnosed in two external assessments earlier today, committed here one commit
later, and catchable by reading fifteen lines at the call site.
WHAT IS CORRECT. _speculative_subjects maps token -> SET OF CLAIMING SUBJECTS.
A token is evicted exactly when the subject that actually claimed it is
reviewed; discarding an absent claimant is a no-op, so a neighbour's marker
survives. Token derivation moved out of the call site into
_speculative_index_tokens, because ownership is what the call site cannot see —
doing the split out there made every token its own claimant, which was the bug.
Tests rewritten to call the cache exactly as §10b does. The lead pin
(test_coherent_proposal_does_not_unmark_an_unrelated_subject) was confirmed to
FAIL against the refcount design before this landed, so it discriminates rather
than merely passing. 13 pins green; the 8 pre-existing lifecycle pins unchanged.
SUITE MEMBERSHIP, resolved rather than deferred. The file pinning this behavior
sat in NO curated suite, reachable only through `full` — which is why the
regression was invisible (G-7's mechanism, on the pin that would have caught
it). Registered now in BOTH TEST_SUITES["smoke"] and smoke.yml, per the #136
precedent that an unregistered pin runs nowhere. Verified the local/CI delta is
unchanged at exactly ten files (local 24, CI 14, CI-only 0), so this is
compatible with all three R-14 options and settles none of them; PR-4 still
owns the standing ten.
Cleanup as found: three unambiguously dead imports removed from pipeline.py
(hashlib, RatificationOutcome, RatifiedIntent). ruff 4 -> 1; the survivor is a
pre-existing structural E402, left alone.
Ratification note unchanged: this changes served surfaces — markers appear on
turns where they previously vanished. Direction is toward disclosure and it
restores ADR-0021 §Articulation's stated intent rather than extending it.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
The speculative-marker cache seeds a subject AND each >=4-char token split from
it, so two independent proposals can claim the same token ("wisdom" from both
`wisdom` and `practical wisdom`). `_forget_speculative_subject` popped the token
outright on COHERENT promotion — regardless of which proposal had seeded it — so
promoting one proposal stripped the marker from another that was still
unreviewed. Unreviewed material was then served WITHOUT the "(speculative, not
yet reviewed)" prefix: the expensive failure direction, on the surface whose
whole job is telling reviewed knowledge from unreviewed.
_speculative_subjects becomes OrderedDict[str, int] — the value is a reference
count. Seeding increments and refreshes LRU position; promotion decrements and
evicts at zero. Seeding and eviction walk the same source list so counts
balance; where they disagree the count stays positive and the subject keeps its
marker, which is the honest direction to fail. An unmatched release floors at
removal and never goes negative, so it cannot borrow against a later proposal's
claim. The LRU cap survives unchanged as a size valve.
Four new pins in tests/test_speculative_subject_lifecycle.py, ALL FOUR OBSERVED
RED BEFORE GREEN — sibling token evicted, marker lost at the served surface,
repeated teaching cleared by one promotion, and count underflow. The eight
pre-existing lifecycle pins pass unchanged.
Also H-8e — expand_relation_closure's docstring claimed "Cycle: if path would
revisit a node, skip", describing a check the code does not perform. The code
was right: termination is structural (monotone growth over a finite triple set),
and a path-revisit skip would REFUSE SOUND transitive derivations, since a
witness path that revisits a node still proves a true fact. The docstring now
says that, and records that `path` is the witnessing step rather than full
ancestry — extending it would move operator_invocation and therefore trace_hash,
so it is deliberately not done here.
Registers updated: H-13 marked FIXED with the ratification note (this changes
served surfaces toward disclosure, restoring ADR-0021 §Articulation's stated
intent rather than extending it); H-8e marked CORRECTED.
Found while fixing it, recorded not acted on: tests/test_speculative_subject_
lifecycle.py — the file pinning this exact behavior — is in NO curated suite,
reachable only through `full`. G-7's mechanism, caught on the pin that would
have caught the defect. Assignment left to PR-4; adding it to TEST_SUITES
["smoke"] alone would widen N-3's ten-file delta and prejudge R-14.
Pre-existing ruff findings in pipeline.py (3 unused imports, 1 E402) confirmed
present before this change and left alone — module-top, not adjacent to the edit.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
Two external first-principles assessments of CORE were verified against code
line-by-line at 8f0ccfe3. What survived is entered; what failed is recorded in
the audit's negative space with the refuting file named, per the register's own
discipline.
Entered:
H-13 — the speculative-marker cache over-evicts shared tokens: promoting one
proposal to COHERENT pops tokens regardless of owner (pipeline.py:736-739,
:1045-1055), so a sibling proposal's still-SPECULATIVE material is served
without its marker. Honesty regression, served direction. Remedy: refcounted
cache, failing test first. (The source assessment described this seam with
the direction inverted — under-eviction; the code over-evicts. The check
that caught the inversion is what confirmed the bug.)
H-14 — _run_turn is 759 lines (pipeline.py:189-947) under a thin-wrapper
docstring, with finalize_turn_trace_hash x2 / finalize_turn_surface x1
back-stamps as the measured symptom. H-4's sibling: refactor ADR, PR-10's
byte-identical-trace bar, and the learn stage named stateful.
H-8e — inductive_closure.py's docstring claims a path-based cycle skip the
code does not perform (:106 vs :189-193) — and the code is the right side:
the monotone fixed point needs no path check, and adding one would refuse
sound derivations. path records only (a,b,c) for longer chains (:198), so
trace provenance is incomplete. Docstring fix is mechanical; extending path
changes operator_invocation bytes -> trace_hash, held to the trace-change bar.
PR-9 widened — the logos-authority bare except (pipeline.py:555) and the
three OOV-probe swallows (:605-657) join the count-the-swallow scope; same
remedy shape (record repr(e), keep the backstop), explicitly not the
narrow-catch alternative.
R-4 packet addition — first collapse candidate: composed_surface +
transitive_surface + transitive_max_depth -> chain_depth:int, justified by
ADR-0083's own docstring ladder; ruled in-lineage, not ad-hoc.
R-12a note — lived_life.json staleness recorded as a question; the proposed
PID-attestation fix rejected (refuses every reboot resume; PID reuse).
Cleared, with the refuting file named: EpistemicState compression
(vault/store.py, policy.py's ADR-0206 partition), the EpistemicState x
NormativeClearance product type (erases refusal evidence — H-3 rebuilt as a
type), PID attestation, path-based cycle skipping, the depth-enrichment cache
(pack_resolver.py:19 already lru_caches the cost), and the unified reasoning
engine as-refactor (trace_hash change + capability widening + discards the
ROBDD engine one merge after ADR-0265 ratified denial).
Register ranges updated to G-1..G-21 / H-1..H-14. Docs only; no runtime
behavior, no flag, no test changed.
[Verification]: uv run pytest tests/test_adr_index.py
tests/test_adr_status_governance.py tests/test_adr_0252_s5_blindness.py -q —
327 passed at this tree.
Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wcw2pnMBwyvmNyQg4uPEt4
The branch was re-created with SSH signatures before landing, which changed my
commit SHAs. Four citations of the pre-registration commit — the verdict's
header and closing note, G-1, and the plan's Track A section — pointed at
dfc394d2, which is not in main's history. They now point at 299c92be, which is.
Same tree, same message, same ordering, signature added. The claim the verdict
rests on is unchanged and checkable at 299c92be: that commit contains the
thresholds and no results. Recorded as a note in the verdict document rather
than swapped silently, because a document citing a commit absent from its own
repository's history is precisely the failure mode its §1 catalogues — and this
one would have been self-inflicted.
Docs only.
[Verification]: uv run pytest tests/test_adr_index.py
tests/test_adr_status_governance.py tests/test_adr_0252_s5_blindness.py -q —
327 passed.
Track A of docs/assessment/50-execution-plan.md §6, executed under its own
protocol. Criterion pre-registered at dfc394d2 (that commit carries the
thresholds as importable constants and NO results); this commit carries the run.
VERDICT: NO-GO. Full credit by ADR-0252 §5.4's own terms.
docs/research/sme-experiment-verdict-797ebad5.md
evals/structure_mapping/adr0252_s5/results/report-797ebad5.json
deterministic_digest b3d9d27592e213104c51bcace7415fd819d722a1b06f7d0a5ca65bd5a234e4ec
53 cases, 1378 pairs, 0.0% non-convergence — no exception contributed to any
number. RS-A (structure-only) separates perfectly, AUC 1.0000, margin +0.0867,
and fails structure-sensitivity. RS-B (attribute-bearing) fails all three.
The mechanism, measured rather than argued: the similarity quotient that would
deliver attribute-invariance is the same quotient that annihilates structural
contrast. An add-vs-subtract minimal pair — one entity, identical numbers, one
relation kind changed — aligns at residual exactly 0.0, its two configurations
being related by a PROPER rotation about e1. Sweeping the attribute weight
(committed as diagnostic_sweep), structure-sensitivity fails at every setting,
and every regime where the SME property survives is a regime where attributes
contribute nothing: AUC 1.00 -> 0.97 -> 0.83 -> 0.69 as they begin to matter.
Scope stated narrowly: this refutes H1 for embeddings that encode role-structure
as point positions aligned by conformal Procrustes under similarity. The
argument is about the quotient, so it generalises across that class; it does not
refute every Cl(4,1) representation, and it does not touch the symbolic
structure-mapping lane already in evals/structure_mapping/.
Two findings that were on nobody's list:
N-8 — the experiment was NOT unrun. rnd/sme-experiment-v2 @ 96e5f468 is titled
"Verdict: GO" and rnd/structure-mapping-experiment @ fc9d0c14 carries an earlier
one. Neither survives inspection: attempt 1 leaked the S1-S4 label into the
embedding; attempt 2 was blind and then took its separability from
`except ValueError: res = 1000.0` — a solver exception counted as a distance —
on a corpus 46/51 of which is outside holdout_dev/v1 with nothing marked, with
duplicate graphs, colliding ids, and an extractor that was never committed. The
register, the plan and the ADR all read the same absence and inherited the same
error for nine days. A branch tip is not a record.
G-21 — the math reader returns a selected graph for 5 of 500 holdout_dev/v1
cases (1.0%), all one skeleton; 24/150 on the public lane. That is why §5.1's
four-structure corpus is not extractable, and it is a sharper measurement of the
comprehension frontier than G-3's construction count.
Registers updated: G-1 carries the verdict, H-10 confirmed and discharged, G-21
added, plan §0 gains N-8, Track A closed, synthesis frontier 2 rewritten.
Pins: tests/test_adr_0252_s5_blindness.py, each observed red before green
(label leak and duplicate-graph sabotage both caught).
Off-serving — evals/structure_mapping/adr0252_s5/ is imported by no serving
path, emits no answers, and changes no flag.
[Verification]: 797ebad5 + this branch — `uv run core test --suite smoke -q`
641 passed in 199.76s; `uv run ruff check` clean; report regenerated after lint
with an unchanged digest.
Track A, per 50-execution-plan.md §6: "state the GO/NO-GO criterion in writing
BEFORE running (the ADR's §5 conformance bar, not a criterion chosen after
seeing results)." This commit carries the criterion and the code and NO results.
Why the protocol is load-bearing here rather than ceremonial: the §5 experiment
is not unrun. It has been run twice and returned GO twice, on unmerged branches,
and neither run's rule was binding.
- rnd/structure-mapping-experiment @ fc9d0c14 — disowned by its own successor:
the S1-S4 label entered the embedding, so Procrustes ran on identical
matrices per structure.
- rnd/sme-experiment-v2 @ 96e5f468 — blind, and then measured separability
with `except ValueError: res = 1000.0`. Its GO test cross_mean > 2*same_mean
is satisfied by that sentinel alone. A solver exception is not a distance.
Instrument characterised before use, on dummy clouds: conformal_procrustes has
two branches. The (5,K) Kabsch+Umeyama path never raises and is invariant to
rotation, translation AND uniform scale (identical 1.8e-16, rotated 7.8e-16,
translated 3.9e-15, scaled x2.5 1.4e-14) while separating non-similar
deformation (5.9e-1) and unrelated clouds (1.1e0). The mixed versor/point
sequence path performs field conjugacy and raises. Attempt 2 used the second;
this uses the first, and treats a raise as non-convergence — excluded, reported
as a rate, never counted as separation.
Corpus fact that forced the design, measured at 797ebad5: the serving reader
returns a selected graph for 5 of the 500 holdout_dev/v1 cases (1.0%), all with
the same skeleton. §5.1's four-structure corpus is not extractable from
holdout_dev/v1 today, so §5.1's own escape hatch is used and MARKED — 5 real
cases, 48 synthetic from four declared templates. A verdict from this corpus
speaks about the geometry, not about coverage; §5.5's compression estimate is
not answerable at 1.0% coverage and is not attempted.
Off-serving. Imported by no serving path; no answers emitted.
PR-1's amendments were derived at ed06dd64. main has since moved (#140, #138),
so every amended claim was re-measured against code at 797ebad5 before landing:
- TEST_SUITES["full"] == ("tests/",) (N-1) holds
- 5 tests/test_l10_*.py, none marked quarantine or slow (N-2) holds
- smoke suite 23 files vs smoke.yml 13, same ten-file delta,
and zero CI-only files (local is strictly the superset) (N-3) holds
- RuntimeConfig: 32 booleans, 28 off, the same 4 on (N-7) holds
- CONTINUOUS_LIFE_CONFIG_FLAGS still omits accrual (N-6) holds
- ADR-0264 §4.1 self-supersession banner in place (N-5) holds
- 18 resolve_promotable_* organs in generate/derivation/ (H-8b) holds
One correction to the amendment itself. H-8d said CONTINUOUS_LIFE_CONFIG_FLAGS
"contains neither claim's subject." It contains one of them: consolidate_
determinations IS forced. The false clause in both docstrings is the one naming
accrue_realized_knowledge as co-enabled, and a third comment on persist_session_
state makes the same shape of claim and is true. The entry now says that, and
says it was caught on re-verification — the same standard §0 holds itself to.
Docs only. No runtime behavior, no flag, no test changed.
scripts/hooks/pre-push describes its own step 1 as "the smoke suite — exact
CI-gate parity" while running 23 files against the CI gate's 13. That is the
second in-repo assertion of the parity H-12 falsifies, and it is inside the
tooling that enforces the protocol.
The corrections derived in 50-execution-plan.md §0, applied where the wrong
claims live. Every amendment is marked [AMENDED N-x] inline so the register
carries its own correction history rather than quietly reading correct.
Gap register — four entries amended:
- G-5 the soak RAN and passed (5000 beats, reboot at 2500, all four gates,
aed273b1). Retitled: what is owed is the committed artifact, the pinned
deterministic_digest, and a cadence — not the run. Both prior claims
withdrawn with their causes: "no suite contains any l10 test" was a
scan artifact (TEST_SUITES["full"] is a directory), and the pins
execute twice a day (post-merge + nightly cron).
- G-7 respecified. The proposed meta-check was already satisfied and would
have shipped green proving nothing — a hollow gate in the entry
proposing to abolish hollow gates. Now two pins: gate parity, and
curated-suite membership EXCLUDING full.
- G-8 28 default-off flags of 32 RuntimeConfig booleans, not seventeen.
- G-10 the 16-premise cap was removed by ADR-0264 R5 on 2026-07-26; the ADR
carries a self-supersession banner directly under the heading Phase 4
quoted. No engineering prerequisite remains — R-8 alone gates it.
- G-19 note: the 21/25 exposure is already pinned in test_volume_honesty.py,
so the work is applying the demotions, not discovering them.
Hindrance audit:
- H-12 added — the local smoke suite (23) and smoke.yml (13) have diverged by
ten files. Counter-evidence stated first: AGENTS.md makes the local run
the merge gate and it is the superset, so nothing is unguarded. What is
real is one comment claiming an equality that does not hold, and the
push that skipped the local gate.
- H-8 gains instance (d), inside the code: two flag docstrings assert a
production profile CONTINUOUS_LIFE_CONFIG_FLAGS does not have.
- H-1 gains the already-pinned note.
Also: correction note on the 2026-07-25 verification doc (a corrective
document that introduced the error at H-8c), and a staleness banner on
MIND-PHYSICS-BLUEPRINT.md pointing at the taxonomy that dissolved it.
The two ratified-ADR amendments (ADR-0146, ADR-0252) are NOT here — they are
drafted in full in 50-rulings.md R-12 and land on that ruling, because editing
a ratified ADR is Shay's authority even when the edit changes no decision.
Docs only. No runtime behavior, no flag, no test changed.
scripts/hooks/pre-push — the automation of the AGENTS.md local-first protocol
— opens by describing its own step 1 as "the smoke suite — exact CI-gate
parity". It runs 23 files; the CI gate runs 13. That is the second in-repo
assertion of a parity that does not hold, the first being core/cli_test.py's
audio block, and it is the strongest evidence available that the drift was
never intended rather than a deliberate narrowing.
Phase 6 of the holistic assessment: what follows from it, and what it needs
ruled before anything else can be sequenced.
`50-execution-plan.md` — five waves (scrub/rule → delete → simplify/enforce →
accelerate evidence loops → automate last) plus five frontier tracks, covering
every G-1…G-20 and H-1…H-12 entry, with four hard dependency gates and nine
risks stated before they are incurred.
`50-rulings.md` — R-1…R-14. One decision per section: the evidence verified at
ed06dd64, the options enumerated, a recommendation with its reasoning, and the
exact diff that follows from each choice, so ruling costs one word. Nothing in
it recommends reversing a ratified decision; R-12 amends two ADRs' records
without touching either decision.
§0 of the plan carries SEVEN corrections to the assessment, all found by
reading code, workflows and ADR supersession banners — none by re-reading
documents. Four change planned work:
- N-3 (new) the local smoke suite (23 files) and smoke.yml (13) have diverged
by 10 files. Severity is LOWER than this plan's first draft stated: AGENTS.md
makes the in-worktree run the merge gate and CI secondary observability, and
the local gate is the superset — so nothing is unguarded. What is real is one
in-code comment claiming an equality that does not hold, and the push that
skipped the local gate, for which CI is the only automatic check.
- N-4 the L10 soak HAS been run — 5000 beats, reboot at 2500, all four gates
pass (contract.md, landed aed273b1). G-5's "never produced an artifact" is
wrong. The debt is the ceremony: no committed report, no pinned
deterministic_digest (the contract's own unmet closing instruction), no
cadence. Prose with no pinned digest is testimony, not evidence.
- N-5 G-10's blocker does not exist. ADR-0264 §4.1 carries a self-supersession
banner: R5 removed the 16-premise cap on 2026-07-26, and four bands would
earn SERVE the moment a ledger is sealed. PR-13 is WITHDRAWN and the whole
throughput frontier reduces to one ruling (R-8).
- N-7 the flag count is 28 of 32 RuntimeConfig booleans, not seventeen.
Plus N-1 (the orphan scan artifact: TEST_SUITES["full"] is a directory),
N-2 (the L10 pins run twice a day, post-merge and nightly), and N-6 (H-8 has a
fourth instance inside the code — two flag docstrings assert a production
profile the production profile does not have, which moves R-3 decisively
toward "incomplete flag set").
Docs only. No runtime behavior, no flag, no test changed.
40-assessment.md — the verdict in one paragraph (skeleton real, discipline
exceptional, the two deepest commitments unproven in opposite directions);
the cognitive cycle stage-by-stage with the honest sentence for each; what
is excellent, stated as findings; the macro picture (built-and-dark,
capability outruns proof, the record decays faster than the code); THE
FIVE FRONTIERS (the reading, the verdict, the chooser, the proof of life,
the throughput — everything else on the registers is hygiene, enforcement,
or ceremony); the recommended attack order in five waves sequenced by the
mastery algorithm (scrub/rule -> delete -> simplify/enforce -> accelerate
evidence loops -> automate last, with the warning that an always-on
process consolidating an empty set is "garbage at high speed" and CORE
came within one flag of it); the charter's four questions answered
directly; and the maintenance contract that keeps this directory from
becoming the next dead instrument.
README.md — the directory index and reading order.
The assessment is complete: 21 documents, four phases, three
self-corrections, every claim SHA-stamped at 8927c563.
30-gap-register.md — CORE's first LIVE gap register since docs/gaps.md
closed its 26th entry (supersession proposed for ruling). 20 entries in
four tiers, each with evidence, deciding authority, and leverage rank:
- Tier A (frontier-blocking): G-1 the unrun ADR-0252 §5 experiment
(leverage 1 — authorized, scaffolded, verdict-less; the paradigm governs
ALL future comprehension); G-2 the #138 fabrications (pre-labeled
measured-and-pinned, held for ADR + ratification); G-3 reader 19 vs
writer 1739; G-4 CR-2 no-chooser (the AGI-grade conceptual absence);
G-5 L10 proof debt (no artifact, no suite, no cadence); G-6 the
half-gated lived learning loop.
- Tier B (enforcement debt): G-7 orphaned-pin meta-check (the highest-
leverage single mechanical change); G-8 flag-default register; G-9 three
doctrine laws without verified failing pins; G-10 curriculum scoping +
absent ledger; G-11 identity authorization bar unstated.
- Tier C (one-line rulings): CR-3 efferent, CR-4 temporal stance, CR-1
attention ADR, the daemon's owning ADR.
- Tier D (latent/carried): aspect-arm defect class (generate/templates.py
:79, unreachable today), sensorium entry criterion, curriculum-formation
bypass (unverified carry), Wilson evidence debt, refusal materialisation.
- Plus what is NOT in the register and why (deferred-with-ruling is the
scripture model; benchmarks excluded by the completeness criterion).
31-hindrance-audit.md — 11 entries ranked by leverage, each with fitness
verdict, evidence, proposed better home, deciding authority; decides
nothing. Headliners: H-1 Wilson/replay counting basis (wrong-solution in
the counting, not the gating — 21/25 bands short; the re-count may demote
licenses and that is the mechanism working); H-2 decoration as testimony
(DriveGradientMap, InhibitionMask — deletion per mastery step 2); H-3 the
typed refusal discarded at the public boundary; H-4 extend the resolver's
declared-precedence pattern upstream; H-8 three record/code contradictions
(each one paragraph to fix); H-9 dead instruments standing as if live.
Plus five examined-and-CLEARED candidates (the 18 organs' continued
service is governance working; pure-Python-by-default is measured-correct;
flag-gated conservatism is not hindrance — unregistered darkness is).
Eight component cards in 20-component-cards/: the four zero-subsystem zones
(comprehend-organ, determine-phase, realize-phase, sensorium-falsification)
plus always-on-process, derivation-organs, surface-selection,
attention-allocation. Plus 05-phase3-findings.md and correction banners on
the M6 and M4 layer cards.
CORRECTIONS (C-1..C-5):
- THREE more map live-serving labels demoted to live-internal:
comprehend-organ (a MATH SETUP ROUTER, not chat comprehension — imported
by neither serving entrypoint), determine-phase and realize-phase (both
behind accrue_realized_knowledge=False / idle flags; NO default-config
serving turn touches any of the three).
- Phase 2's M6 card overstated ephemerality: Shape B+ persistence exists
(persist_session_state, daemon-forced, "restored bit-exactly").
- Phase 2's M4 card understated the resolver: surface_resolution.py IS a
declared-precedence resolver for the pipeline seam; only the upstream
composer arms remain ordered branches. Third Door refines to "extend the
existing pattern upstream."
- 34-vs-18 RESOLVED: 18 resolve_promotable_* organs existed AT the ADR-0252
ratification commit; "34" counted modules (~32-33). No consolidation has
occurred; the governing ADR's headline number has no stated basis.
- The 2026-07-25 verification doc's claim that accrual "is enabled by the
production L10 process" is contradicted: CONTINUOUS_LIFE_CONFIG_FLAGS is
exactly {persist_session_state, consolidate_determinations,
strict_identity_continuity}.
NEW FINDINGS (F-6..F-10):
- F-6 THE CONTINUOUS LIFE MAY CONSOLIDATE AN EMPTY SET: the daemon forces
the consolidator ON but not the accruer, and realize_comprehension — the
only turn-path writer of realized facts — sits behind the accrual flag.
The two halves of the lived learning loop are gated by different flags
and only one is forced. Undocumented either way.
- F-7 ADR-0008's Allocation Physics LANDED, mutated, ungoverned: curvature
kernel -> salience adapter -> attention threshold plan -> candidate
intersection, with budget feedback making attention SELF-NARROWING
across a walk. InhibitionMask is decoration. top_k=16/threshold=0.3 have
no recorded derivation. CR-1 closes with a one-page ADR.
- F-8 two load-bearing naming traps (comprehend-organ; realize vs realizer).
- F-9 REALIZE is sound; its ceiling is the reader that feeds it (the #138
fabrication locus — measured/pinned/held, recorded not re-discovered).
- F-10 _accrue_in_turn swallows exceptions into a no-op with no telemetry.
Phase 4 seed list consolidated: 8 hindrance candidates, 10 gap candidates,
each with its deciding authority. Method note recorded: three phases of
corrections, and nothing in the chain was caught by re-reading documents.
Fills the taxonomy with evidence. Every liveness label re-verified at
8927c563 rather than inherited from the 2026-06-09 system map. Three
inherited claims did not survive, and every one of them came from a
document while every falsification came from reading code.
CORRECTIONS:
- Phase 0 Finding 0-C was WRONG. The L11 always-on process IS BUILT:
chat/always_on.py::run_continuous, chat/always_on_daemon.py::run_daemon
(single-instance lock, SIGINT/SIGTERM stop, load-time identity guard),
CLI `core always-on`. Landed 2026-06-14 — five days AFTER the system-map
snapshot that declared no forever entrypoint exists. The real finding is
sharper: built, with a complete falsifiable soak harness, never run to a
recorded artifact, and enforced by NO SUITE AT ALL.
- CR-1 attention/allocation is not a missing layer but a live undocumented
one: generate/stream.py:255-263 runs SalienceOperator -> AttentionOperator
and gates every generation step; use_salience defaults True. It owns the
~73%-of-turn hot path and has no ADR, no card, no layer in any ratified
articulation. (generate/salience.py COMPOSES core.physics.salience — not
a duplicate; that false lead is recorded, not propagated.)
- CR-2's components exist: DriveGradientMap and ExertionMeter are both
constructed in chat/runtime.py. DriveGradientMap is NEVER READ — textbook
decoration by the sabotage test. ExertionMeter runs but feeds telemetry
only; fatigue gates no decision. The claim survives and sharpens: CORE has
drive objects and no chooser.
STAGE-COVERAGE AUDIT: 7 of 9 covered; non-text ingest uncovered (59
sensorium modules reach no serving path); the cycle's runner claimed-only.
CROSS-CUTTING: F-1 built-and-off is the dominant pattern (17 flags default
False, only deduction_serving ratified ON) with no register of the set or
what would flip it. F-2 enforcement lags capability — orphaned pins are
undetectable because suite tuples are hand-curated. F-3 formation's
six-boundary trust standard is not applied at M2, the surface facing
untrusted user text. F-4 ADR/code contradictions are load-bearing (the
daemon has no ADR while ADR-0146 rejected the daemon shape; ADR-0252's
"34 surface organs" does not reproduce — 18 found). F-5 what is excellent,
stated plainly: the typed learning boundary, selection-not-rewrite, the
non-hardening invariant, fail-closed lane shapes.
No layer is wrong-solution. Two candidates deferred to Phase 4 with
evidence: Wilson/replay independence in licensing, DriveGradientMap.
Fixes the decomposition Phases 2-3 fill and Phase 4 audits against, per
Pillar I: criteria before judgment. Five decisions recorded explicitly:
- D1 The five unreconciled articulations answer five DIFFERENT questions
(what for / what happens today / by what mechanism / judged how / built
where) and become axes of one taxonomy rather than competitors. Functional
axis = the AGENTS.md north star; structural axis = 7 macro layers (M0
Substrate, M1 Knowledge & Memory, M2 Afferent, M3 Comprehension &
Reasoning, M4 Expression & Serving, M5 Learning & Growth, M6 Continuity &
Process) + 2 cross-cuts (MG Governance, MV Verification), grouping all 33
system-map zones with straddles recorded honestly. Mind-physics blueprint
dispositioned element-by-element (identity/compositional/renderer =
superseded by stronger implementations; Allocation Physics and
DriveGradientMap = orphans -> Candidate Register).
- D2 System-map schema EXTENDED, not replaced: 7-value liveness kept;
four missing dimensions added as orthogonal fields — evidence (with
would-fail-if-absent sabotage bit), capacity (designed/measured/ceilings),
design-vs-build split, fitness (6-value: fit/strained/misplaced/
wrong-solution/superseded-in-place/undetermined). Liveness ⊥ fitness:
the 34 organs are live-serving + superseded-in-place, a state no merged
enum could express.
- D3 Completeness criteria fixed: coverage = evidence-bearing ownership of
a functional stage; deferred-with-ruling ≠ missing; pins must run.
- D4 Candidate Register: CR-1 attention/allocation (the ~73%-of-turn-time
hot path has no owner), CR-2 agenda/drive (mechanisms without a chooser;
L10 risks landing as a heartbeat with nothing to want), CR-3 efferent
action (needs an explicit deferred-vs-out-of-telos ruling), CR-4 temporal
self-location (determinism vs lived time, undesigned). Considered-and-not-
registered recorded (sociality, affect, embodiment, self-governance-as-
layer) so the boundary is visibly examined.
- D5 Assessment output is COMMITTED (reviewable, versioned, visible);
system map stays local as a regeneratable navigation index. Staleness
handled by verified_at SHA stamps on every card.
Phase 2 work order: nine layer cards, priority M6 > M3 > M4 > M5; re-verify
every map liveness label; run the stage-coverage audit with evidence; the
four zero-subsystem zones scoped honestly and queued first for Phase 3.
Opens the holistic macro→micro assessment of CORE's cognitive-cycle design
and implementation fulfillment. Read-only; no runtime behavior changes.
Method is governed by docs/conceptualizing_engineering_mastery.md (committed
here because the assessment cites it as its method): Semantic Rigor,
Mechanical Sympathy, Third Door; delete before optimizing; automate last.
Phase 0 findings:
- 0-A CORE carries FIVE unreconciled macro articulations of its own cognitive
cycle (AGENTS.md north star; AGENTS.md live path; the 2026-05-12 draft
three-physics-layer blueprint; ratified ADR-0252's five paradigm stages; the
system map's 10-band L0-L11 layering). The blueprint's allocation/attention
layer (Salience/Attention/Inhibition) exists in no other articulation.
- 0-B .system-map/ is the best existing macro→micro artifact and is 48 days
stale, gitignored/local-only, and carries four zero-subsystem zones — three
of which sit on the serving path.
- 0-C ADR-0252 §5 — the experiment gating the governing paradigm's build — has
never returned a verdict (two unmerged rnd/ worktrees). The 34 surface organs
the ADR diagnoses as the error keep serving. L10 blocks all 7 OPEN W-* items;
field + T1 vault are discarded on exit, so "one continuous life" is today
many short lives sharing a checkpoint.
- 0-D No live gap register: docs/gaps.md is 26/26 closed, the liveness ratchet
is L10-blocked and stale, docs/analysis/ is a chronological archive.
- 0-E 333 ADRs, flat numbering, status not reliably machine-extractable;
INDEX-by-domain is 21 files behind; the Blueprint reserved 0254-0261 for
intents never materialised while those numbers were minted for other work.
- 0-F Measured: CGA is ~73% of turn time via cga_inner/geometric_product
(~34k calls/turn), NOT versor_condition (0.22%). Rust backend off by default,
parity question open and blocked. MLX in no runtime path.
- 0-G The 2026-07-25 verification precedent falsified ~1/3 of an external
blueprint's items by reading code — the bar for every downstream phase.
Phase 1 receives the system map's 17-field proto schema and 7-value liveness
vocabulary, plus the four dimensions it lacks (evidence pointer, capacity,
design-vs-built, fitness judgment) and five decisions to make explicitly.
PR #138's fabrication findings are recorded as measured-and-pinned, held for
ADR + ratification — never re-discovered, never fixed here.
Phase 5 item 1 asked for the overlap to be sized "by measuring the reader's
construction set against the writer's rather than by growing corpora blindly".
New lane `evals/construction_inventory` does that: it sweeps the writer's whole
parameter space (every RhetoricalMove x IntentTag x predicate x quantifier x
tense x aspect, both public entry points, 32292 cells), quotients it by the
reader's OWN function-word skeleton, and comprehends each construction under
three vocabularies.
writer constructions 1739
reader constructions 19 (mint-site AST-guarded)
overlap, faithful 6
accepts but MIS-READS 22
refuses 1711
vocabulary-dependent 0
Faithfulness, not acceptance, is the criterion — and that reverses two claims
in the plan's §6 RESULT, using a metric that was already on the page:
* g_read_rate is 1/293 but **g_args_rate is 0.0**. The one surface that "reads",
`all molecules are defined as compounds`, is comprehended as
`subset(molecule, defined_as_compound)` — the reader chunked the writer's verb
phrase into a class name. It accepted; it did not comprehend.
* "one construction wide" was wrong both ways: zero on that corpus, six over the
writer's actual output space. A corpus cannot report an inventory's size.
Three findings re-order Phase 5 item 1:
1. The reader FABRICATES on 22 constructions — neither reads nor refuses.
`every dog is a mammal` -> member(every_dog, mammal);
`furthermore, all dogs are mammals` -> asserted(furthermore).
Both are ordinary user English, not writer artefacts. Root cause: _RESERVED
lacks the function words the writer emits, and _parse_propositional accepts
any single token as a fact.
2. It reaches SERVED output. deduction_surface recites
`Given: furthermore; p implies q; p.` — a premise the user never stated — and
chat/runtime.py realizes declarative turns into the held self, so a
fabricated atom is vault-writable. Widening the inventory first would widen
the fabrication surface with it.
3. ADR-0265's defect class survives inside ADR-0265's designated owner of clause
grammar: the four aspect arms of _inflect_predicate bind `negated` to a
wildcard and never read it, so `dog has been defined as mammal` serves both
the assertion and the denial (10530/16146 points). Unreachable today — no
producer sets aspect — so a loaded gun, not a casualty. It survived because
ADR-0265's invariant is structural (is `negated` threaded?) and cannot see an
arm that receives the flag and ignores it. A behavioural sweep can.
The two fixes are written already, as the mutations that turn the defect pins
red. They are NOT applied here: they change what CORE comprehends from user
input, which is a serving change on the truth path — authorization-gated, ADR
first, on the ADR-0261 §5.1 refuse-don't-drop precedent.
Guards, so the tables cannot rot: reader constructions are pinned to an AST
count of the reader's 10 mint sites; the declared tense/aspect axes are pinned
to _inflect_predicate's match arms; the committed corpus is pinned to the
fillers in use. 14 pins, 13 mutations observed RED.
[Verification]: in-worktree, CPython 3.12.13, uv sync --locked.
deductive 517 (was 503, +14 — count moved, registration confirmed)
smoke 641 (unchanged; the file is registered in `deductive`)
lane SHAs 11/11 match, no pin edited
pyright 0 errors on all new files
mutations 13/13 red, including both fabrication fixes
Serving-path authorization for the negation fix in this PR. The ADR is
Accepted rather than Proposed because the defect it removes was LIVE behind a
shipped flag (realizer_grounded_authority), not latent: CORE served the
affirmative of propositions users denied.
CORE served the affirmative of propositions the user denied. Measured live on
main @ 536d6e55 with realizer_grounded_authority=True:
"evidence does not support truth" -> 'Evidence is verified: what supports truth.'
"evidence supports truth" -> 'Evidence is verified: what supports truth.'
Byte-identical.
Phase 4 (#136) pinned "render_semantic has no negated parameter" as a defect.
Sizing the exposure before acting on it -- the step §7 requires -- showed the
pin was correct and INCOMPLETE. The defect is two drops in series, and the
first is in the graph:
intent.negated parsed from the user's words, always has been
GraphNode NO FIELD FOR IT <-- first drop
ground_graph rebuilds nodes field-by-field; nothing to carry
depth enrichment rebuilds nodes field-by-field; nothing to carry
plan_articulation cannot carry what the node does not hold
realize_semantic never read step.negated
render_semantic no parameter for it <-- second drop
Five separate constructors, each silently defaulting to the affirmative.
Why every gate was green
------------------------
On the default config the ungrounded realizer emits "...", _is_useful_surface
rejects it, and the runtime echo wins the resolver -- and the echo contains the
user's own "does not". The truth path was correct BY ACCIDENT. ADR-0088 Phase B
grounds the graph first, which is exactly when the realizer's surface becomes
useful enough to win. So the defect sat behind a shipped flag, invisible to
every property-of-one-surface test, because nothing compared a denial to its
assertion.
Delivered
---------
- GraphNode.negated, threaded intent -> graph -> ground -> enrich -> step ->
surface. Serialized ONLY when True, so every pre-existing as_dict and every
trace_hash folded from one stays byte-identical -- which is why no lane pin
moves.
- Clause grammar delegated to its one owner. Four of the eight intent
"templates" were never frames; they were plain clauses. They now call
render_step. Writing a second negation implementation in semantic_templates
was the tempting fix and is rejected: Phase 2A spent a unit giving every
linguistic fact one owner, and that design rebuilds the disease one level up.
- Frames that keep a finite verb (VERIFICATION, PROCEDURE, COMPARISON) get an
explicit negated form. RECALL is a speech act with no proposition to deny, so
it falls back to the clause path rather than drop the denial (ADR-0261 §5.1).
Measured
--------
delegation on affirmatives 192/192 byte-identical
serving realizer, all corpora 85/347 -> 109/347 (+24 = the denials)
feature-bearing bucket 49/214 -> 73/214
CONTROL (nothing droppable) 33/33 -> 33/33 unchanged
multi-node (clause joining) 3/100 -> 3/100 unchanged, out of scope
lane pins 11/11 byte-identical, none edited
The control staying 33/33 and multi-node staying 3/100 is the evidence that
this moved the denials and nothing else.
The exhaustive control earned its keep immediately
--------------------------------------------------
It found FIVE more intents serving a denial as its own assertion --
TRANSITIVE_QUERY, FRAME_TRANSFER, NARRATIVE, EXAMPLE, DEDUCTION -- because an
intent with no frame fell back to the UNKNOWN *template* (which cannot say
"not") rather than the UNKNOWN *clause* (which can). The default is now the
capable path, so the next intent added inherits correctness. Found by the
control, not by inspection.
Mutation -- every link reverted individually
--------------------------------------------
baseline 19 pass
graph_from_intent drops intent.negated 8 FAIL
plan_articulation drops node.negated 6 FAIL
ground_graph drops it on rebuild 2 FAIL
realize_semantic stops passing step.negated 6 FAIL
unframed intents fall back to the template 1 FAIL
pipeline depth-enrichment drops it 1 FAIL
The last row was GREEN on the first run -- my fix there was unguarded. That is
what added the structural invariant: every GraphNode(...) built on the serving
path must NAME `negated` or be recorded in an allowlist with a reason. The
defect was five constructors; a per-site test must be written per site, and a
site added without one is invisible. Only recognition/connector.py is exempt
(an EpistemicNode has no polarity to carry).
Four Phase 4 pins are revised, not relaxed -- #136's M2 mutation ("render_semantic
GAINS a negated parameter -> FAIL") has now happened for real, and forced the
deliberate revision it was built to force.
Registered in `smoke` in the same PR that creates it, per the #136 finding that
an unregistered pin runs nowhere.
Still unexpressed, deliberately: quantifier, tense, aspect. NO PRODUCER sets
them anywhere on the serving path, so threading them would be machinery with no
caller. render_step already handles all three the moment a producer exists.
[Verification]: in-worktree on CPython 3.12.13 with `uv sync --locked` --
smoke 641 (was 621), deductive 503, lane pins 11/11 with no pin edited.
Phase 4 of the grammar-unification arc, resolved by option (b) of the plan.
The problem: 149 green fluency cases scored `realize_target`, and
`core/cognition/pipeline.py` never calls it -- it calls `realize_semantic`.
`english_fluency_ood` reported 117/117 + 39/39 for a function that does not
speak, for the whole life of the lane.
The plan framed this as "which realizer is better". Reading the source says it
is not a quality question. `render_semantic`'s signature is
`(intent, subject, predicate, obj, secondary, language, root)` -- no `negated`,
no `quantifier`, no `tense`, no `aspect` -- and `realize_semantic` never reads
them off the step. So the serving writer cannot express content the
ArticulationStep is carrying:
negated=False -> 'Knowledge is defined as opinion.'
negated=True -> 'Knowledge is defined as opinion.'
It serves the AFFIRMATIVE of a negated proposition. That is the ADR-0261 §5.1
family, not a fluency defect. Pinned here as a defect; NOT fixed, because
fixing it changes served output and belongs to Shay.
Measured -- identical contract, all seven scored corpora
--------------------------------------------------------
bucket n realize_target realize_semantic
feature-bearing, single node 214 207 49
no features, multi-node 100 100 3
no features, single node (CONTROL) 33 33 33
total 347 340 85
The control is what makes the rest mean anything. Every corpus hardcodes
IntentTag.UNKNOWN, so "the serving writer scores badly" could have been an
artifact of never giving it a real intent. On the 33 cases carrying nothing it
cannot express, the two are IDENTICAL -- so the gap on the other 314 is the
dropped features and the clause joining, not the intent.
Delivered
---------
- `grammatical_coverage/runner.py` reports `serving_accuracy` beside
`accuracy`; `english_fluency_ood` delegates to that run_lane and gains it
for free. The realizer is a parameter now instead of a hardcoded import.
- `tests/test_phase4_realizer_resolution.py` -- the control, the decomposition,
the negation defect pin, and the §6 evidence.
- The claim is restated at both places it was made: the lane docstring and
`grammar_roundtrip/contract.md`.
Zero served bytes change. No serving authorization needed.
§6 -- the plan's pre-commitment was wrong, and is corrected rather than
quietly edited
------------------------------------------------------------------------
§6 forks on read_rate: risen => Phase 5 diversity; near-zero => §1.8 graph-model
mismatch => ADR. It pre-committed to the second. Neither is what the
measurement says.
g_read_rate went to 1/293. The unblocked case was blocked by a one-line WRITER
defect (predicate-nominal object agreement, #135), not by §1.8. The other 292:
no_template_match 289 reader has no SUBJ-VERB-OBJ template at all
unknown_morphology 2 prepositional objects (reserved_word_in_np)
unsupported_negation 1 reader has no negated-categorical template
Every one is the reader declining a CONSTRUCTION, not a projection disagreeing
about a graph it parsed. Where a construction is in both inventories the round
trip closes exactly. So the barrier is the OVERLAP of the two construction
inventories, currently one construction wide -- Phase 5's item 1, tractable,
not an ADR-scale model decision.
§1.6 read "uniform no_template_match" as evidence FOR the type mismatch. It is
not: no_template_match is a coverage fact, and the corpus was 289/293 bare
transitives, a construction the reader never claimed to read. The measurement
was mostly reporting the corpus's composition.
And a finding in my own stack
------------------------------
`tests/test_realizer_quantifier_agreement.py` lived ONLY in the `cognition`
suite, which is not on the AGENTS.md pre-push gate. So every pin Phases 3 and 4
added to it -- including the invariant covering all twelve inflection branches
-- ran in NO gate. That is why smoke stayed at 621 across two PRs that added 13
tests between them, and it is the same silent-red shape the smoke list already
calls out for test_adr_index.py.
Registered into `deductive`, which now runs 504 instead of 406.
Mutation
--------
baseline 12 pass
serving metric computed with realize_target 4 FAIL
render_semantic GAINS a `negated` parameter 2 FAIL
realize_semantic delegated to realize_target 6 FAIL
Row 2 matters most: if someone FIXES the negation defect, the pin forces a
deliberate revision instead of passing silently.
Still open, still Shay's: option (a), promoting realize_target to the serving
path. It now has a price tag -- 340/347 over 85/347 and the ability to say
"not", against a move in every surface hash and whatever the Shadow Coherence
Gate ruling in core/cognition/surface_resolution.py was protecting.
[Verification]: in-worktree on CPython 3.12.13 with `uv sync --locked` --
smoke 621, deductive 504 (was 406; +12 new Phase 4 pins, +86 previously
ungated), lane pins 11/11 unchanged, no pin edited.
Claude
