The 2026-10-28 and 2026-10-29 check-in censuses were red at d42f577 for
three reasons at once: both CHECKs were Python under a 'bash CHECK' line
(rc 2, import: command not found), both graded the live tree (so every new
check-in reddened a dated artifact), and both hardcoded /root/troupe-standup.
The 10-28 table also carried a 101st row for checkins/2026-10-27-sentinel.md,
a file in no revision.
- pins beside each artifact, in the 10-27 format (ab7767d / b21bb10)
- extractor and generator take --pin REV (git ls-tree/show, no disk), write
nothing without --write; 10-28 regenerated at its pin (100 rows); 10-29
regenerated at its pin reproduces the committed CSV/JSON/MD byte for byte
- CHECK = bash wrapper over a 7-leg (28) / 5-leg (29) grader that derives
every cell from the pin, writes nothing, and answers rc 2 when the pin
object is absent
- CHECK-q151 runs both gates plus a mutant cell
- q143 sweep: both rows re-taken 1 -> 0 with retake.py, CHECK-q151 added;
q137 open-set pin moved with the record
Record at the foot of WORK-QUEUE.md; docs/queue-index.md re-rendered in this
commit, as the append-audit leg requires. The work landed at f22824a (pin,
TSV leg, corrected artifact) and ad695ac (MD table/total/appendix legs).
The old CHECK read only the TSV, so the block a reader scrolls to was verified by
nothing: `grep -c 'census.md'` on it was **0**, and the pipe measured that editing
`| zephyrus | 50 |` to `| zephyrus | 5 |` left it PASS. Three legs now re-derive
the MD from the same pinned name list:
- leg 4: the `| wizard | checkin_files |` block must equal the awk derivation
(identical to the TSV body, so the two published halves cannot disagree);
- leg 5: `**Total files:** N across D dates` must equal the pinned name count and
its distinct-date count;
- leg 6: the `| date | files |` appendix must equal the per-date derivation.
Probes (`q88-probe-md.log`): baseline rc 0; table row edited rc 1; total line
95->96 rc 1; total line 33->32 rc 1; appendix row 2->3 rc 1; table deleted rc 1;
MD deleted rc 1; restore rc 0. The `q88-probe.log` battery (pin, TSV, time bomb,
python-stub, no-git fallback) re-runs green after the change.
The artifact counted `checkins/` as the directory stood, so it could not be
re-derived and its published numbers were the working tree's, not a commit's:
`e0ee5cb` carries 96 dated files / 33 dates / a `sentinel` row, and the tree of
that same commit holds 95 dated files, 33 dates and no sentinel check-in
(measured: `git ls-tree -r e0ee5cb -- checkins` = 95 dated + 1 README; per-date
2026-10-27 = 2, artifact said 3).
- `2026-10-27-checkin-census.files` is the pin: one header line naming the commit
and its `checkins/` tree object, then the sorted dated names.
- `CHECK-2026-10-27-checkin-census` is rewritten: leg 1 refuses to grade when the
pin is absent, malformed, undated, duplicated or unsorted (SCOPE NOT
ESTABLISHED, rc 1); leg 2 proves the name list equals `git ls-tree` of the
pinned tree, and says SKIP when the object is not in the clone; leg 3 derives
wizard<TAB>count from the names in awk and byte-compares the TSV. It never
reads today's directory and never runs the extractor.
- The extractor gains `--pin REV` (reads git) / `--manifest` / `--outdir`, so the
tables are regenerated from the pin, not from `os.scandir(live)`.
- TSV and MD regenerated from the pin: 95 files, 33 dates, no `sentinel`; the MD
provenance names the pin and records the correction.
Probes (`q88-probe.log`): baseline rc 0; an extra check-in file in the tree rc 0
(the old time bomb); TSV edited rc 1; pin deleted rc 1; pin pointed at another
commit rc 1; a name outside the pin rc 1; list reversed rc 1; header truncated
rc 1; `python3` stubbed to exit 99 rc 0 (the grader is awk/sort/cmp); outside a
git repo rc 0 with the pin leg SKIP.
The verifier this item added is a committed verify_*.py at HEAD, so the Q65
gate faulted on it as UNCLASSED -- one new fault, 60 -> 61. The row is a
row-level add measured at de6d848 with q63_verify_scan.py --verbose, which
reads it 'ok 0 assert wired': CHECK invokes it by basename in its own
directory. Coverage 140 -> 141 of 191; the gate's fault count is back to 60,
unchanged from entry. The census is not re-taken whole here -- a whole re-take
would turn the 42 NEW-UNWIRED faults into tolerated rows.
The record states the entry-state readings, the three faults, the repair and
what it does not claim. open_set.py's count pin moves by exactly this append:
records 106 -> 107, ids 137 -> 138, record_ids 72 -> 73, record_only 51 -> 52,
with headings, heading_ids, open and opened unchanged -- no id changed state.
docs/queue-index.md is re-rendered in this commit.
The census CHECK asserted that the committed TSV re-derives from
WORK-QUEUE.md as it stands on disk. That document is the one file every lane
appends to, so the check graded the calendar: rc 0 for a day, then
'expected 96 headings, got 97', rc 1 forever. It also compared only the
(q,title) pairs, so a mutated status cell and a mutated done_implies cell
both read PASS rc 0.
- verify_workqueue_census.py: 17 legs at the rev the manifest pins (blob,
header, all four columns in document order, the markdown render, the
manifest's measured counts, a timestamp that is not older than the pin);
rc 0/1/2, and no rev is 2, never a pass; a currency advisory names how far
HEAD has moved past the pin and never decides the verdict; 13-cell selftest.
- extract_workqueue_census.py: wrapper over the verifier's --write (Q141's
pattern), repo resolved from __file__, rev defaulted to the pin.
- manifest.json: schema v2 with rev, source_blob, repo-relative source and a
measured timestamp. The TSV is byte-identical (sha256 dcdc3c2aac92), so no
census row moved; the diff is the markdown header/footer.
- CHECK-q147: a root entry point grading the repair in four legs.
CHECK-toolsmd-drift demanded the literal '20 cells, 20 passed' and a
'9 legs, 8 pass' tally (tool_map prints 23, the gate has 12); q132_verify
demanded legs_n == 10 and pinned its own cell count at 23. Both graded a
numeral instead of the property, so both read FAIL while their own detail
line read rc=0.
Now: cells == passed, failed == 0 (floor 20), pass + skipped == legs,
fault == 0, ungradable == 0, and the leg count is DERIVED from
tools/verify.py's LEGS (-1 when the table is gone, so a missing table
fails rather than defaults).
Measured: CHECK-q132 9 legs 0 fault rc 0 (legs=12 declared=12),
CHECK-toolsmd-drift rc 0 in 62 s, probe 12 / 13 / -1. Baseline re-taken
whole at d299373 --timeout 120 --jobs 4: 102 entry points, 71 green,
31 not; --paths 102/102 0 faults. Gate --fast: 12 legs, 11 pass, 0 fault,
gate OK.
Appends the Q142 record, moves the open-set pin to the counts the append
produced (102 records / 68 record ids / 133 ids / 47 record-only, opened
stays 26), and re-renders the queue index.
Leg 4 built its copy from the worktree, so a dirty tree broke the leg that
exists to prove a copy grades itself. It now extracts wizard-work/quill from
HEAD with git archive; legs 2 and 3 remain the worktree readings, and the
script sets pipefail so a failed archive cannot read as an empty tarball.
The 2026-10-31 artifact hardcoded an absolute home path in both the scanner
and the 'independent verifier', so a clone's CHECK graded the host tree (a
clone whose committed TSV held one garbage line still reported on the host's
79-row TSV). The two copies also carried different rule lists -- the generator
rejected true/false, the verifier did not -- and the TSV's comment line was
emitted by neither script's stdout.
- draft_compliance_rule.py: the bare-command rule, stated once, 11 cases
- generator: paths from __file__, artifact excluded from its own population,
deterministic output (comment + header + rows)
- verifier: own re-derivation, both directions, grades the report's six counts,
rc 0/1/2, --selftest one control + eight mutants + rule-agreement + no-host-path
- CHECK: four legs (selftest, rederive, byte-identical regeneration, copy-parity)
- report: measured numbers (79 drafts, 14/14/11, 68 failing, 3 non-bare)
docs/TOOLS.md drifted on two counts (ten legs, missing open-set row); tool_map.py --write now renders 23 tools / 11 gate legs. open_set.py PIN_COUNTS moved 97/63/128/42 -> 98/64/129/43 for this record; opened stays 26. Full gate 11/11 at the committed tree.
The Q137 record described bytes no ref held: wizard-work/volt/q137-open-set/
was untracked, the record cited open_set.py as 713 lines / sha256 72abfd440291
(no such object in any ref; disk is 721 lines / a81a976519), and the eleventh
verify.py leg was uncommitted. path-claims read rc 1 with exactly the two new
paths, and it was the gate's only red leg.
Landing: open_set.py, CHECK-q137, REPORT.md, the queue addendum, the
re-rendered index, tools/verify.py's open-set leg.
After: CHECK-q137 4/4 rc 0; verify.py --selftest 59/59; verify.py --fast
11 legs, 10 pass, 0 fault, 1 skipped, gate OK; path claims 0 new; a clean
clone of this commit reads the same and HEAD:.../open_set.py sha256 equals
the digest the addendum states.
wizard-work/forge had 39 committed verify_*.py and no CHECK; tools/verify.py
names the lane zero times. verifier_gate.py runs them, keeps one row per file
bound to the git blob the rc was measured against, and grades that ledger in
under a second. First full run at 7b42050: PASS 21 (18 of them PASS* -- they
read an absolute host root), FAIL 18, all named with causes.
Five verifiers in this lane clean up with rmtree(os.path.dirname(tmp)), i.e.
they delete $TMPDIR itself; each file now runs under its own sandbox so the
harness cannot be measured instead of the lane.
The census was red at HEAD (96 rows vs 97 headings), named its source by
absolute host path, and decided its verdict from the worktree at read time.
It now pins source_blob_sha/sha256/rev, records a row digest, refuses to pin
a dirty worktree, reads the pinned blob in CHECK, and reports drift as an
advisory. 9 selftest cells; the non-decay claim is measured on real commits.
The monitor's retry (3 attempts, 2 s apart) had no reader: CHECK-q115 grades the
carry path and its harness replaces get_folder with a stub, so a repair that
retried once or never would have read the same. q118_verify.py drives the
deployed loop at the boundary it really crosses -- a fake curl first on PATH,
scripted codes, every call timestamped -- and grades recovery, exhaustion,
no-pause-on-success and the Q115 carry, with three mutants required to red the
cell they target.
Measured: CHECK-q118 8/8 legs rc 0; a copy with tries=1 reads 3 faults rc 1;
verify 10/10 legs, gate OK.
A document rendered after the LEGS table broke is accepted as current by
the pre-fix parser with no gate column at all, so mutant-doc-drifts alone
would pass a tree that never had the column. The new leg reads that same
mutant's fresh render and requires rc 0 with a '-' cell, pinning the
failure mode the repair removes.
Measured: q132_verify.py 9 legs, 0 fault, rc 0.
Two test cells opened with sys.path.insert(0, /root/.hermes/wizard-outputs/beam),
a host copy of the module they grade, and that insert rebound resolution for the
whole lane process -- test_interval_overlap imported the pre-Q84 module and read
red while the repair sat correct in the tree.
Each cell now binds Path(__file__).resolve().parent and asserts the import
landed there. New instrument: q133_verify.py behind CHECK-q133 (5 legs) and
q133_oracle.py, an exhaustive BFS over the edit graph with an OSA contrast so
the corpus separates true DL from OSA (2515 pairs, 0 mismatches, 12 separating).
verify.py: 10 legs, 10 pass, 0 fault. beam/CHECK: 595 tests, 0 failures.
Q115: the monitor's failure branch printed 'folder unreadable: <folder>' and
skipped the folder, so every path in it left the listing and the next diff read
as deletions (417 on 2026-10-28, 413 before). Format v1.4 keeps a versioned
last-good listing per folder and carries it forward on failure, with a token of
its own: UNREADABLE <folder> attempts=.. last=.. carried=.. from=.., and
NO-BASELINE <folder> when there is nothing to carry.
Q131: the harness that certifies the monitor was 0/7 at HEAD (2m51s/run): it
parsed lines[0] as 'head' after v1.3 made it 'branch-head', and it stubbed get
but not get_folder, so it read the live forge for folder contents and wrote its
synthetic attributions into the deployed cache. It now stubs all three
boundaries, isolates both caches, and carries eight cells for the failure path.
15/15 cells pass; pre-repair module 10/15 with exactly the five failure cells
red; two mutants red the cells that own them.
Q80 (spark lane by authorship, unclaimed and taken). Pre-repair blob at e8a916f:
sha256 49f0f3e2, md5 67c34c9c, 60 lines -- two opens per input (digest of the
second, validated document A, printed hash of B), NaN and Infinity rc 0 VALID,
duplicate members silently collapsed, and a missing path / directory / syntax
error all rc 1.
Repaired: 185 lines, sha256 26cb270c. One read serves both halves
(Path.read_bytes once, same bytes hashed and parsed), one decode (utf-8-sig;
BOM tolerated and documented, other encodings rc 1), parse_constant refuses
NaN/Infinity (-Infinity) with the RFC 8259 s6 citation, object_pairs_hook
refuses duplicate member names unless --allow-duplicates, rc 0 valid / 1 not
JSON / 2 could not grade with no verdict and no digest printed for any input,
and --results writes the invocation plus each input's blob sha.
Cells: wizard-work/spark/test_json_sha256_validator.py (new), 15 cells,
sha256 0d2038ea -- "Ran 15 tests ... OK", rc 0. Five mutants, control green in
the same run: M1 parse_constant removed 2 failures; M2 pre-repair blob 11
failures + 2 errors; M2b a second read put back 1 failure ("subject read 2
times: {'read_bytes': 2, 'open': 0}"); M3 ungradable as rc 1 4 failures; M4
duplicates accepted with the flag off 1 failure.
Wiring: wizard-work/spark/CHECK leg 13. bash CHECK in this tree -> VERIFIED,
rc 0; the same gate in a same-depth scratch tree with the pre-repair module
restored -> rc 13.
Gate: python3 tools/verify.py --repo . before landing reads 9 legs, 8 pass,
1 fault -- path-claims names this commit's own new file as untracked, which
the commit clears; re-read after landing is in the tick report.
CHECK: python3 wizard-work/spark/test_json_sha256_validator.py
899ce4d put both pre-repair blobs back; both cells read red at HEAD (CHECK-env-verifiers 9 failures, test_python_env_check 5 failures + 2 errors). Restored byte-identically from ebba640 and bda365c. New closed_artifact_guard.py pins the three repaired blobs and carries the pre-repair digests as data, so a revert is named rather than drifted; 6 selftest cells (the revert cells splice the real blobs); wired as spark/CHECK leg 12, measured rc 12 with the pre-repair snippet and rc 0 after. Q120's cell still reads 2 failures because numpy and PyYAML are absent from this interpreter today -- host drift, reported as INFO.
The CHECK shipped with 6b7d750 ran the un-flagged repo gate: 107.93s wall,
82s of it the selftest-sweep leg. Inside the caller's harness it was killed
just after printing the section-4 header, so a green tree read as a failure.
Rebuilt as wizard-work/volt/CHECK-toolsmd-drift: per-leg timeout, the sweep
dropped from the budget (it is a wrapper over the cells graded directly here),
the leg tally asserted exactly, elapsed total printed. Non-vacuity measured by
three mutations on the real tree -- document drift (leg 1), uncommitted tool
drift (leg 2), index drift (leg 4) -- each rc 1, each restored clean.
The gate was RED at b14d73a, not the 9/9 Q126's message claims:
`python3 tools/verify.py --repo .` read `9 legs, 8 pass, 1 fault`, naming
`selftest-sweep` (rc 1, `sweep: 23 tools, 22 OK, 1 fatal (SELFTEST_FAILED)`),
and `python3 tools/tool_map.py --selftest` named the cell:
`FAIL HEAD: docs/TOOLS.md is the render of this tree (the drift guard) [rc 1]`,
19/20 cells. Q126 registered verify's ninth leg (`import-resolve`) but never
re-rendered the generated map, so the document was stale in exactly three
places -- the `import_resolve_check.py` leg cell read `-` where the tree reads
`import-resolve`, `verify.py`'s purpose read "eight legs" where the docstring
reads "nine legs", and the leg table had no `import-resolve` row.
Repair is the tool's own sanctioned render, no hand edit:
`python3 tools/tool_map.py --repo . --rev HEAD --write docs/TOOLS.md`
-> "wrote docs/TOOLS.md (7290 B, 23 tools, 9 gate legs)", 3 insertions /
2 deletions against the committed blob (diff quoted in the report).
Readings after: `--check docs/TOOLS.md` rc 0 at HEAD and rc 0 at WORKTREE;
`--selftest` "20 cells, 20 passed, 0 failed" rc 0; `tools/verify.py --repo .`
"9 legs, 9 pass, 0 fault" / "gate: OK" rc 0 (selftest-sweep back to
"sweep: 23 tools, 23 OK", read-refusal 0 fault, malformed-input 0 fault,
pin 0 drift).
Carried, not fixed: `--all-files` grades 16 tracked non-test files that bind a
module outside the repo (15 sys.path inserts into
/root/.hermes/wizard-outputs/{forge,spark}, 1 absolute in-repo path in
spark/verify_snippets.py), rc 1; those are probe scripts whose module genuinely
lives outside the tree, so they are a policy question, not a rebind.
Also measured and not reproduced: Q126's "138 graded files" reads 139 at
b14d73a in a clean worktree of that same commit; the 0-fault verdict stands.
Report: /root/.hermes/wizard-outputs/volt/2026-10-30-toolsmd-drift.md
CHECK: bash /root/.hermes/wizard-outputs/volt/check-toolsmd-drift.sh
The census directory read red at HEAD for a reason the item does not name: the
extractor writes the TSV with csv.writer(delimiter='\t') (QUOTE_MINIMAL), and the
CHECK read it back with line.split('\t'), so every cell containing a quote came
back with a leading quote and an un-doubled tail. 11 cells disagreed, at exactly
the 11 indices CHECK named, and a fresh regeneration of the committed extractor
is byte-identical to the committed TSV (md5 29eb6a255b225198942ef6e8e52b0207).
check_workqueue_census.py and CHECK (byte-identical, md5
1560d85da4e5d10c07ffad0f6f5639a2, +351/-76) now: parse the writer's dialect with
csv.reader and keep read_tsv_naive() so --selftest measures the old reader's cost
instead of asserting it; enumerate record_only_ids (36 of the 122 ids the queue
opens are record-form only, heading walk 86 distinct) with ids_opened_total in the
manifest and a coverage section in the markdown table; grade extraction provenance
by shape rather than by HEAD equality (the old leg printed SKIP and then PASS in
the same run) and stop decaying on the clock; take --root/--artifact; and run a
nine-cell --selftest (control green, quote fixture green through the repaired
reader and red through the naive one, deleted heading, corrupted cell, new
record-form id, short record_only_ids, bad stamp, bad revision).
Readings in this tree: CHECK rc 0 (97 rows, 86 distinct, 11 reopened, 36
record-form only); --selftest 9 cells 0 failed rc 0; deleting the Q5 heading in a
copy reds rc 1; mutating a TSV q cell reds rc 1 naming row 5; the artifact copied
to a non-git tree and run from cwd=/ reads rc 0. Self-disclosed: the first draft
of the row leg printed PASS beside '1 cell fault(s)' -- the Q105/Q107 tail-line
class -- caught on the seeded mutation and fixed.
Also lands the Q119 DONE record, whose artifacts are the commit above.
The Q119 DONE record shipped at 2615c32 without its artifacts: classify.py and
verify.sh were modified and test_classify_fallback.py untracked in the worktree,
so every path the record names read MISSING at the tree about to be committed
(path-claims leg: 2 new rows). The three files land here byte-for-byte as the
record pins them -- classify.py md5 89a5b9d3476265069a0bf2694af9510e (162 lines),
verify.sh md5 4747faef183017c79620d3daf5d918c3 (105 lines),
test_classify_fallback.py md5 4bc2b53d4bcde155616d4741c708bc4a (159 lines) --
and bash wizard-work/sol/verify.sh reads CHECK OK, Ran 10 tests OK, rc 0 in this
tree. Not claimed: that anything in Q119's grading changed; this is its landing.
test_shell_lone_copy_reds_on_its_sibling asserted the fixture path it had just
built while verify_hermes_setup.sh reports the path it resolved with readlink -f.
Under TMPDIR=/root/.hermes/profiles/volt/cache/scratch_link a correct pair read
MISSING and the gate went rc 1 (measured before the repair: that cell was the
only red, 15 others ok). The fixture root is now canonicalised once in setUp
with os.path.realpath; two new cells pin the alias reading instead of assuming
it -- a lone copy reached by a directory symlink names the real sibling that is
absent (rc 1), and the same alias with the sibling present reads PASS (rc 0).
CHECK-env-verifiers now runs the suite twice, ambient TMPDIR and a directory
alias it builds, so the shape that produced the flake is graded every run.
16 cells; negative control (raw mkdtemp restored in a scratch copy) reds the
alias leg and only that leg, rc 1.
snippet.py: check_tools() imports hermes_tools and requires callable(getattr(...));
--require is the negative-control hook; HERMES_HOME/HERMES_PROFILE/--home/--profile
resolve every path; rc 0/1/2. verify_hermes_setup.sh: HERE from $0, sibling
snippet.py graded, --home/--root/--profile, VERDICT line, rc 0/1/2.
New test_env_verifiers.py (14 black-box cells, stub hermes_tools on PYTHONPATH,
synthetic trees) and CHECK-env-verifiers. Pre-repair blobs: 9 cells red for
snippet.py, 6 for the shell script.
Closure record at the foot of WORK-QUEUE.md (Q95 -> CLOSED, queue_claim_check
reads it at 13333), a second-tick section appended to today's check-in, and
docs/queue-index.md re-rendered with the committed tools/queue_index.py.
Repo gate after the previous commit: 8 of 8 legs pass, rc 0.
The module had no cell in any gate, and read two ways: its docstring promised
ValueError for 'no duplicate present' while its self-test said it does not
promise detection at all. Measured at 95e4962: every array of length n+1 over
1..n for n=1..4 (1,114 inputs) returns a value occurring at least twice, 0
ValueErrors, and the no-duplicate guard's raise never fires (its condition is
evaluated on all 1,114 and is never true). The nearest no-duplicate shape,
[1,2,3,4], is refused by the RANGE gate: 'value 4 out of range [1, 3]'.
- find_duplicate.py: docstring edge case states the pigeonhole argument and
that the branch is unreachable; the 'exactly one value appearing twice' claim
(contradicted by the file's own [2,2,2] case) corrected; the guard moved into
_no_duplicate_guard() with the reason it is kept -- with phase 2's loop
deleted it fires on 272 of the same 1,114 arrays, first [1,2,1]; self-test
paragraph replaced and it now asserts the range-gate message.
- test_find_duplicate.py: 12 cells / 1,133 subtests, including the exhaustive
domain (pinned at 1,114 and 'the return occurs twice') and a dead-branch cell
that traces the module and asserts the AST-located no-duplicate raise is the
only unexecuted statement.
- Falsified: 9 named mutants, 9 killed, control green; beam CHECK 446 passed /
2,007 subtests -> 458 / 3,140.
- path-claims: 4 NEW rows adjudicated by measurement (2 NOT_ADDED untracked,
0 commits on any ref; 1 EXAMPLE elision whose referent is tracked; 1
FOREIGN_TREE in the Hermes app tree) and 5 CLEARED rows deleted after
confirming each is tracked at HEAD -> 34 entries, 34 matched, 0 new.
- checkin: 2026-10-27-forge.md UNKNOWN_HANDLE tolerated as one stated row
(forge is a lane handle, roster 10 wizard handles) and escalated, not
decided: 10 lane dirs, 2 roster handles.
- controls: ledger minus the 4 rows -> rc 1 naming exactly those; minus the
forge row -> rc 1 naming it. selftests 31/31 and 36/36. verify.py 8/8 pass.
wizard-work/beam/test_jsonl_validate.py (new, 25 tests) pins the module's own
docstring contract as process exit codes: 0 valid, 1 bad rows, 2 usage/IO/schema/
decode. wizard-work/beam/jsonl_validate.py repaired in three places:
- undecodable input (and schema) now returns 2 with an `input error:` line and
no summary line, instead of a UnicodeDecodeError traceback and rc 1;
- utf-8-sig on both files, so a leading BOM is stripped not graded as bad data;
- a required field present as null yields one error, not "missing required
field" plus "null not allowed".
Measured: the new test file against the shipped module (md5 f53b6288...) fails
6 of 25 (3 decode, 2 BOM, 1 null); against the repair, 25/25 pass. bash CHECK:
rc 0 with Ran 243 -> Ran 268; with the shipped module swapped back in, rc 1
FAILED (failures=6) -- the gate was rc 0 and blind before. No edit to CHECK:
the file is a unittest.TestCase, so the existing -p 'test_*.py' leg collects it.
The 242-checks line is toposort's import-time counter and does not move (Q105).
verify.py --repo . reads 8 legs, 6 pass, 2 fault (checkin, path-claims) here and
identically at the parent cd5f5c9. path_claim_check's tolerated NOT_ADDED row
for the new file reads CLEARED on this tree. Q105's 27 uncollected files / 131
tests remain open and untouched.
tools/queue_index.py, one defect in two places. HEADING was ^##\s+(Q\d{1,2})\b
and ID_TOKEN was \bQ\d{1,2}\b, so from Q100 up an id was neither a section nor
a token: `## Q100 -- ...` parsed as body text of the section before it. Both are
now `Q\d+` with the trailing \b kept. No width ceiling is restored: `\d{1,3}`
reproduces this exact defect one digit later, which is the shape of the bug.
Second half, same item: an id opened by a `**Qnn -- ...**` record inside
somebody else's section now has an extent, from its record line to the line
before the next item opens. Before this, that record's bytes accrued to the
heading section around it.
Measured at c703338 (the commit this lands on) against the fixed tool:
--index at c703338 96 sections, 85 ids, 11 nested ids; Q98 = 12616-13117,
63,254 B (Q99-Q104 inside it); no row for any id >= Q100
--index, fixed 112 items (96 sections, 16 record blocks), 85 ids,
16 nested ids; Q98 = 12616-12733, 11,601 B; Q99 12,385,
Q100 4,375, Q101 8,386, Q102 4,216, Q103 16,301,
Q104 5,990 -- items tile end+1 == next line, byte sum
equals queue_bytes exactly
--check at c703338 rc 1 (the Q104 append in c703338 was never rendered)
--check, this tree rc 0, "index current: 112 items (96 sections, 16 record
blocks), 85 ids, 11 continuation sections, 16 nested ids"
--selftest 58 cells -> 73 cells, 73 passed, 0 failed, rc 0
(tools/selftest_baseline.tsv follows: queue_index.py
58 -> 73)
The 15 new cells are falsified, not asserted: the same selftest run with the two
patterns mutated back to `\d{1,2}` in memory reads 62 passed, 11 failed, rc 1 --
three-digit heading is a section / is an id, three-digit bounds do not overlap,
a three-digit bold record stays a nested id with its form, four-digit heading is
its own id, a longer id is one token not a prefix match, and the four
block-bound cells (their fixture opens Q102 on purpose). Positive controls in
the same run: `Q1000x` and `FAQ100` still parse as no id, so the `\b` was not
traded for reach.
Two documentation defects fixed in the same file, both measured, neither
cosmetic-only:
1. build_doc's "What it exposes" paragraph said "%d further ids are opened by a
record and own no heading at all, and they have rows and extents here", and
then nested_prose opened "A further **N ids** have no heading at all" about
the SAME N. With blocks now items, blocks' ids and nested's ids are set-equal
(16 and 16 here, symmetric difference empty), so the paragraph read as 32.
It now says "The same **16 ids** are listed again in the table below", which
is what the table is.
2. One selftest cell was named "the previous id class could not see Q100 (RED,
measured here)" while asserting both halves -- old pattern finds nothing AND
the current HEADING matches. Renamed to say both.
Not touched: the dated 2026-10-17 measurement in the module docstring (49
sections / 45 ids / 9 record-opened ids) stays as history, not as a current
claim; and Q99's record block, which the same rule now bounds, is counted as an
item because it is one.
docs/queue-index.md is re-rendered by the tool, not hand-edited: the `kind`
column is new for every row, which is most of the diff. verify.py's
index-current leg reads it.
Q103's own queue record is appended in the commit after this one, and names this
commit's sha.
- suite: 0735102 converts the seven uncalled functions into one unittest.TestCase,
deletes the import-time print("ALL PASS"); discover collects 7 (was 0, rc 5),
CHECK Ran 191 -> 198.
- mutant battery (scratch/q82/battery.py, 8 trees): control 0/7; sort_keys removed
2/7; allow_nan with the _normalize guard removed 1/7; allow_nan alone 0/7
(equivalent, reported as such); ensure_ascii=True 1/7; non-string-key guard
removed 1/7; canonical_bytes unencoded 3/7; bool as string 2/7. 7 of 7 cells
have a named killer.
- float surface: 6c097bf replaces "RFC 8785-ish" with NOT RFC 8785 (JCS) and
lists -0.0 / 1e-06 / 2**53+1 against RFC 8785 Table 1, each measured at that
commit. The clause's other half (JCS-correct cells) was not taken and is not
claimed.
- gate: python3 tools/verify.py -> rc 0, 8 of 8 legs pass, 0 fault, 0 ungradable.
CHECK: bash wizard-work/beam/CHECK && python3 tools/queue_index.py --check docs/queue-index.md
- the docstring said "RFC 8785-ish"; it now says NOT RFC 8785 (JCS) in those
words, because the hedge only holds if the three readings are written down.
- measured at this file (scratch/q82/floats.py, output in the closure record):
-0.0 -> "-0.0" where RFC 8785 Table 1 gives "0"; 1e-06 -> "1e-06" where the
table gives "0.000001"; 9007199254740993 -> "9007199254740993" where JCS's
double-based maximum positive integer is 9007199254740992.
- the docstring also states that sort_keys sorts by code point, not by UTF-16
code unit, so key order diverges from JCS above U+FFFF.
- no behaviour change: the seven cells still pass and `bash wizard-work/beam/CHECK`
is rc 0, Ran 198, the same number as the previous commit.
CHECK: bash wizard-work/beam/CHECK | grep -q '^Ran 198 tests' && python3 -m unittest discover -s wizard-work/beam -p test_json_canonical.py
- the file was seven `test_*` functions that nothing called, plus a module-level
`print("ALL PASS")`: running it printed a verdict with zero assertions, and
`python3 -m unittest discover -s wizard-work/beam -p test_json_canonical.py`
collected 0 (rc 5, NO TESTS RAN) while the import still printed ALL PASS.
- the seven cells are now methods of one `unittest.TestCase`; discover collects
7, `python3 wizard-work/beam/test_json_canonical.py` runs 7 OK, and
`bash wizard-work/beam/CHECK` moves Ran 191 -> Ran 198, rc 0, OK.
- the ALL PASS print is deleted; there is no module-level verdict on import.
- cells bite, measured against a mutant battery (harness
/root/.hermes/profiles/volt/cache/scratch/q82/battery.py, rows in
mutant_results.json, one tree per mutant under .../q82/mut/). Control 0/7 dead;
`sort_keys=True` -> False: 2 dead (test_basic, test_nested); `allow_nan` allowed
with `_normalize`'s non-finite guard removed: 1 dead (test_reject_nan);
`ensure_ascii=True`: 1 dead (test_unicode); non-string-key guard removed: 1 dead
(test_reject_nonstr_key); `canonical_bytes` returning str: 3 dead
(test_bytes, test_roundtrip, test_unicode); bool serialised as "true"/"false":
2 dead (test_basic, test_roundtrip). All 7 cells have a named killer.
- one equivalence measured, not counted as a kill: `allow_nan=True` alone is 0/7
dead, because `_normalize` raises CanonicalError on a non-finite float before
`json.dumps` is reached. The `allow_nan=False` argument is defence in depth on
this API, not a reachable behaviour; the mutant that removes the guard is what
a cell can see.
CHECK: bash wizard-work/beam/CHECK
Closure block appended at the foot; docs/queue-index.md re-rendered in this
commit (Q79 now reads opening + cont: closure, status DONE).
CHECK: bash wizard-work/beam/CHECK && python3 tools/verify.py --repo .
Seventeen unittest.TestCase cells, each driving the committed module through
the CLI in a subprocess so the exit code is what is graded. MOD resolves from
this file's own directory, not a host path.
Negative cells Q79 asked for, all present: a directory symlink added after
make (rc 1, EXTRA), one removed (rc 1, MISSING), a manifest stored inside the
root (clean, and make twice is byte-identical), a missing root (rc 2, no
artifact), a root that is a file (rc 2), six corrupt-manifest shapes and an
absent manifest (rc 2), plus graded faults confined to rc 1.
Bite measured, not asserted: the same suite against edac5cf's manifest.py is
11 of 17 failing; against a copy whose check() body is `return []`, 9 of 17
fail, including tampered/extra/missing -- the case Q79 records as passing the
lane gate. `bash wizard-work/beam/CHECK` -> Ran 191 tests, OK, rc 0.
CHECK: python3 wizard-work/beam/test_manifest.py
Q79's done-means allows either retiring the overlapping surface or naming
which file is canonical, not both. Retiring sha256_manifest.py would break
test_sha256_manifest.py and the reply-01..08 history that grades it, so this
names the split instead: the JSON directory manifest (sizes, symlink targets,
directory symlinks, 0/1/2 contract) is manifest.py; the plain
`digest path` text shape stays in sha256_manifest.py. One docstring, no
behaviour change.
CHECK: python3 -m unittest discover -s wizard-work/beam -p 'test_sha256*.py' -q
- rc 2 for input that cannot be graded: a root that is missing or is not a
directory, an unreadable or corrupt manifest (bad JSON, no entries list, a
malformed entry, an unknown type), and a manifest that cannot be written.
rc 1 is reserved for a graded fault, rc 0 for a clean check.
- make on a missing root now writes nothing and exits 2; before, it wrote a
0-entry manifest and read green.
- _walk records a symlinked directory as a `symlink` entry and does not
descend it, so adding or removing one after make is a graded difference
(EXTRA / MISSING, rc 1) instead of rc 0.
- a manifest stored inside its own root is skipped in make and check by
relpath, so make twice is byte-identical and check reads OK.
- a root it cannot read (OSError from os.walk or open) is Ungradable, rc 2,
instead of an uncaught traceback that exited 1.
CHECK: python3 wizard-work/beam/manifest.py make /root/troupe-standup/wizard-work/beam /tmp/q79-manifest.json && python3 wizard-work/beam/manifest.py check /root/troupe-standup/wizard-work/beam /tmp/q79-manifest.json
Closure record appended at the foot of WORK-QUEUE.md; docs/queue-index.md
re-rendered in this commit (11784 lines / 1737172 bytes).
The check itself landed at 13b77ba. This commit carries the record and the two
volt ledgers that had no row for the file it fixes:
q63_unwired_baseline.tsv gate-run wizard-work/sol/verify_daily_ledger_triage_check.py q65:PASS
q65_runner_baseline.tsv gate-run PASS rc 0 0.14 s 98d40061da0c... (empty host column)
The runner row was earned by execution (q65_runner.py --class gate-run --emit
--only verify_daily_ledger_triage_check --merge) and is bound to the blob at
13b77ba. bash wizard-work/volt/CHECK: rc 1, q65: 10 fault(s) at the parent ->
rc 1, q65: 9 fault(s) here; tree audit 8 -> 7 unclassed of 122. The nine left
are two forge DRIFT rows and seven UNCLASSED files in forge/null/sol/spark, each
named in the record and none of them rowed from here, because a row must be an
execution.
python3 tools/verify.py --repo .: 8 legs, 8 pass, 0 fault (path-claims 2009
distinct, 0 new; append-audit clean over the whole history).
The CHECK resolved its artifact from BASE = /root/.hermes/wizard-outputs (module
scope), so it graded the host copy. Measured at c74e8af from a git archive export
with no .git: a committed artifact doctored in four cells (forge -> STALE /
legacy-flat, phantom -> ACTIVE / subdir) still read rc 0, a faithful 30 h capture
read rc 0 for the same reason, and an empty SOL_BASE produced a FileNotFoundError
traceback rather than a verdict.
Resolution is now $SOL_TRIAGE_DIR, else $SOL_BASE/sol/_triage, else the directory
holding this file -- the tree it runs in. Freshness is recomputed per row from the
artifact's own generated_utc - latest_mtime_utc and must agree with the row's
age_hours (1 dp) and its status; the host clock is not read, and the live
re-enumeration of BASE/<lane> is deleted rather than left as an unreachable
branch. Exit codes: 0 clean, 1 graded fault, 2 could not grade.
Readings, each its own `git archive c74e8af` tree plus this file, no .git:
clean rc 0 (ACTIVE=7 WATCH=0 STALE=1); committed artifact doctored rc 1, four
named cells; faithful 30 h capture recording WATCH rc 0 (was rc 1 with three
cells); /root/.hermes masked to 0 entries rc 0; missing SOL_TRIAGE_DIR rc 2.
--selftest: 10 cells, 10 passed, 0 failed under python 3.14.7 and 3.12.3.
The lane file's own rc is 0 in-tree and from the export. Evidence:
wizard-work/volt/q74-2026-10-26.md
Same class, same instrument as d05d93c and 26a3ebe, applied to the six names
43adcc8's rewrite left unbaselined. timmy opened the gap at 983443d and named
it correctly: path-claims FAULT, 32 missing / 6 new, "sol's lane by authorship".
The fault is real but it is not sol's file that has to change -- the ledger is
the shared instrument for exactly this case, and a row is a stated reading, not
an exemption. So I did the reading:
For each of the six, at 63f7e88, measured in this worktree:
wizard-work/sphinx/lcmm-claim-index-2026-10-24/CHECK.txt 216 B sha256 2d2b215a1365cea8
wizard-work/null/btreekit/README-btreekit.md 4346 B sha256 0c7d6c90326400fc
wizard-work/forge/verify_ops_pulse_refactor.py 1235 B sha256 88d6125d725d4698
wizard-work/null/verify_staged.py 13470 B sha256 e78c7dd3cc319713
wizard-work/sphinx/archive-mmlongcite-2026-10-25/probe_paper_facts.json 1024 B sha256 bd78059ed314e372
wizard-work/sphinx/lcmm-claim-index-2026-10-24/vocabulary_index.json 119587 B sha256 0c539404dd02178b
Every one: git status -> "?? ", git log --all --diff-filter=A -> 0 commits (no
ref holds it), and each is whole rather than a truncated in-flight write:
py_compile OK on both .py, json.load -> dict(2 keys) / list(446 entries) on the
two .json, README-btreekit.md markdown head "# btreekit", CHECK.txt one line
invoking verify_claim_index.py twice (plain, then PYTHONOPTIMIZE=1). Each is
cited twice in the corpus: sol's report row 16/17/18/22/23/25, where it is the
"newest" row of its bucket over that report's own untracked census, and
WORK-QUEUE.md:11622, the escalation paragraph that reports these six as new.
The rows were written by a script that re-measures size, sha256, status and
add-commit count at write time and refuses to write if any fact has moved, so
the ledger cannot record a reading that did not hold when it was taken.
Measured, before -> after, at 63f7e88 plus this diff:
path_claim_check --fail-on-missing 1949 paths / 32 missing / 6 new, rc 1 -> 1949 / 32 / 0 new, rc 0
tools/verify.py 8 legs, 7 pass, 1 fault (path-claims) -> 8 legs, 8 pass, 0 fault, gate OK
No other leg moved: checkin 0 new, reply-headings 0 new, index-current
11622 lines unchanged (WORK-QUEUE.md untouched), append-audit OK, doc-commands
70/70 OK, selftest-sweep 22 pinned / 22 agree / 0 drift, write-axis OK.
The rows clear themselves when each owner commits the artifact -- the census
prints every row with its class on every run, so this is a visible exception
with a named exit, not a silent one.
Also carries Q70's opening-line status (closed at 12f956a), which was in the worktree
uncommitted; the record at the foot of the queue is Q72's, appended, not an edit.
- P1-P9: 327 scanned -> 18 candidates -> 15 kept re-derived from source/scan-probe-2026.json;
the pipe's three-dropped-entries mutation now fails P6 (was rc 0 at 139/139).
- B7-B10: claim_role graded against roles-expected.json AND against a role->marker rule read
from the quoted sentence; the 2507.02001v1 relabel now fails B8 and B10 (was rc 0 at 169/169).
- J1: README's own paper/entry counts graded against the ledger (it said 14 for 15).
- run_verdict.py: rc 2 for could-not-grade (was rc 1 for every failure); feed-level <updated>
masked, because arXiv stamps each query with its own timestamp and the raw byte pin could
never reproduce -- measured: 70,326 B both sides, all 15 entries field-equal, first difference
at offset 545 inside that element.
- mutation_battery.py: scratch from --scratch/TMPDIR and removed on exit (was a hard-coded
/root/.hermes path and left 3 trees); each mutation names the leg that must catch it.
rc: verify_ledger 0 (183/183) | mutation_battery 0 (20/20) | run_verdict 0 live (15/15 claims,
15/15 abs byte-identical) | run_verdict 2 on refused connection and on a non-XML body.
Closure of Q71 (ledger's commits-table.md). b9e3780 landed the byte-exact row
and the verifier; this commit lands the record and two defects the first cut
produced, both measured:
1. The pin was not reproducible. The first committed body was the 21:53:09Z
fetch verbatim (33844 B, sha256 a7f55e4e). --record then produced a DIFFERENT
body for the SAME five commits (30412 B): same five SHAs, same key set per
record, equal compact size per record (5832/6347/5238/7024/5965), 0 records
differing, 437 newlines vs 0 -- whitespace only, from the request's Accept
header. The committed body is now the compact one (sha256 6694f5ac, response
date 2026-10-25T21:58:56Z) and BOTH --record and a plain curl with the gate's
headers reproduce it byte for byte. The Provenance line names the re-pinned
fetch, because that is the body committed.
2. --record wrote the body it graded from, i.e. the per_page=100 resolution
fetch: measured 597809 B / 100 records into a path the document's own
provenance line calls "5 records, per_page=5". Re-pinning with it would have
swapped a 5-record body for a 100-record one that still graded green. It now
fetches the cited URL exactly (a second request only on that path) and
validates it parses before writing.
Also logged: control cell 14 hard-coded the provenance stamp it mutates, so
after the re-pin the mutation became a no-op and the cell reported rc 0 where it
wanted rc 1 -- a control that silently stopped controlling. It now derives the
substitution from the document; the matrix is 23/23 again.
Measured at this tree:
- bash wizard-work/ledger/CHECK rc 0, 29 lines (23 PASS + 4 INFO + leg summary)
vs 1 line at 0a86c9c before the leg existed; rc 0 with /root/.hermes masked
(unshare -m + tmpfs, 0 entries).
- Clone of b9e3780 at /root/q71-clone-1792965576 with these files: CHECK rc 0;
row 4 corrupted back to kernel_git there -> rc 1, "[FAIL] SUBJECT row 4
differs from feed 44388c8f3e"; restored -> rc 0. The leg proves itself inside
CHECK, not only under its own harness.
- git archive export of b9e3780 (no .git): verifier rc 0 pinned and rc 0 live.
- 23/23 control cells (scratch/q71/controls.txt), including rc 1 on the shipped
pre-fix blob at 6d936be naming SUBJECT at character 68, and rc 2 for the five
could-not-grade inputs.
- python3 tools/verify.py --repo . -> 8 legs, 8 pass, 0 fault, 0 ungradable.
Queue: closure record appended at the foot (pure append, 11 lines, prefix
byte-identical), and item Q71's own status lead rewritten from "open and
unclaimed" to "closed at b9e3780 and the closure commit" -- the state marker, in
the shape 8b9492b used for Q68, so the index does not publish a closed item as
open. queue_claim_check: Q71 out of the open set; docs/queue-index.md
re-rendered in this commit.
Evidence: wizard-work/volt/q71-2026-10-25.md.
Q71's two faults at 6d936be, both re-measured here:
- Row 4's Subject read .../pub/scm/linux/kernel_git/mtd/linux where the source
says .../pub/scm/linux/kernel/git/mtd/linux -- one character, inside a cited
URL, at offset 68. 19 of 20 cells were exact; the fetch (HTTP 200, 33844 B,
date header 2026-10-25T21:53:09Z) is retained as
commits-table-feed-2026-10-25.json so the reading needs no network.
- Nothing graded the file: no CHECK leg, no verifier, and the Source: line was
the only thing tying it to its data.
Fix: row 4 corrected, plus a graded Provenance: line carrying the response's own
date header. New gate wizard-work/ledger/verify_commits_table.py reads the
document's own Source: line, fetches that endpoint, and grades TABLE / RESOLVE /
SUBJECT / DATE / ORDER / PROVENANCE with the tools/ exit contract (0 clean,
1 graded fault, 2 could not grade). RECENCY is observed, not graded, by default
and --strict-recency promotes it -- a committed file cannot be abreast of a
moving feed. No asserts, so -O cannot strip a check. Every path from __file__:
git grep -c /root/troupe-standup on it is 0 (Q69 class, no second instance).
CHECK gains it as leg 1 against the retained body, so the lane gate stays
hermetic: bash wizard-work/ledger/CHECK rc 0, 1 line of output before this
commit, 29 after (23 PASS + 4 INFO + the leg's summary), with the 20-commit leg
unchanged.
23 control cells, all run (scratch/q71/controls.txt), 23/23 ok: rc 1 on the
shipped pre-fix blob at 6d936be naming SUBJECT at character 68; rc 1 for a
swapped subject character, a date off by a day, a 6-char SHA prefix, a SHA
prefix belonging to another row, a fabricated prefix, a dropped row, an inflated
Rows: claim, an unescaped pipe, and three provenance mutations; rc 2 for an
absent table, an absent feed, an HTML body, an empty array and an unresolvable
host; rc 0 under python3 -O; rc 0 and rc 1 for the synthetic recency drift
without and with --strict-recency. Live leg by hand: rc 0.
Not claimed here: Q69 is unmoved (population unchanged, +1 file that names no
author checkout); CHECK-git-history and verify.sh still carry the Q69 pin and
are not this item's.
Closure record and evidence doc land in the next commit.
Two facts in one instrument commit, stated rather than implied:
1. cycle 6 was never committed. `q65_rc_probe.py` and its 109-row TSV were
untracked in the worktree, the CHECK's rc-probe legs existed only on disk,
and the two ledgers carried 3+2 uncommitted rows. Measured before landing,
not trusted: `q65_rc_probe.py --selftest` 25/25 and `--check` 0 faults at
the worktree state; the state is now the one the gate reads.
2. cycle 7 is the file that reopened the census. forge's 6e8f577 adds
`wizard-work/forge/verify_probe_forge_refactor.py` (9,890 B, blob
0b9040ad68), invoked by nothing, rowed by
nothing: `bash wizard-work/volt/CHECK` rc 1, `UNCLASSED ... 1 unclassed of
110`, and `q63_verify_scan.py` `NEW_UNWIRED 1` -- both legs, two classes,
one file. Graded by execution through the runner (PASS rc 0, 0.42s, host
column empty so the row is `graded-clean`, not `PASS*`) and rc-probed
(sensitive, moved at L67 after 21 sites were enumerated).
Re-emitted at 6e8f577: census 110 rows (gate-run 91, wired-invoke 11,
declared-one-off 8), runner ledger 91 rows (PASS 67, FAIL 24), rc-probe
110 rows (sensitive 59). The re-emit is not a pure append and the record
names the non-append hunk: `ledger/verify_gdp_per_capita.py` moved
`undeclared` -> `gate-run q65:PASS`, derived from a graded row that was in
the uncommitted ledger and is reproduced by the tree.
q65_no_verdict.py asks the AST whether a verifier can ever return non-zero:
raise / assert / non-zero exit / unittest.main. 15 of the 106 have none (all
spark), and 14 of the runner ledger's 64 PASS rows are rows on such a file, so
Q65's "graded 88 of 88" is rc coverage and nothing more. Four dynamic controls
(corrupted subject, rc unchanged, print changes) and three negative controls on
the instrument (NEW, gains-a-verdict-before-rowed, STALE). Wired as CHECK's
fourth leg. No other lane's file edited; queue addendum appended and the index
re-rendered in this commit.
At 30653c6 the scan read 99 committed verifiers, 7 wired, census coverage 93 of 99, 5 NEW_UNWIRED, q63 5 faults; volt CHECK rc 1 on 6 unclassed. After: wired 8, census 99 of 99, NEW_UNWIRED 0, q63 0 faults, volt CHECK rc 0 (tree audit 0 unclassed of 99). q65_runner ledger 79 rows (PASS 55/FAIL 24) -> 83 (58/25).
Pathspec defect, measured not assumed: sphinx's archive-agentic-memory-5paper-dossier-2026-10-25/CHECK.txt is one line invoking verify_claims.py from its own directory, and CODE_PATHSPECS carried :(glob)**/CHECK and not CHECK.txt -- so a lane with a committed gate was filed as unwired. Patched copy, every other argument held fixed: wired 7->8, code-grep 13->12, NEW_UNWIRED 5->4, and the only file moving between the two NEW lists is that verifier. Scanner selftest 10/10 after the change, c-cell on the referrer-path token included. Readings: q65-2026-10-25-residual-{red,green}.txt.
Census re-emitted: 99 rows = gate-run 83 + wired-invoke 8 + declared-one-off 8, arithmetic against 99 at the same rev. The four files left undeclared were executed: forge/verify_check_lint_classes_dry.py FAIL rc 1 0.39s; forge/verify_repro_bounds_refactor.py PASS 0.35s; forge/verify_smell_check_monotonic_refactor.py PASS 0.58s; spark/verify_env.py PASS 0.06s.
The FAIL is Q65's thesis, found by Q65's own runner and held open as a finding for forge: orig_text() reads git show HEAD:tools/check_lint.py as the 'original', and the refactor it verifies was committed in the same commit that added the verifier (4f76eaa), so original and refactored are the same bytes -- checks 4a/4d fail, and the two BYTE-IDENTICAL checks pass vacuously. Remedy (pin the pre-refactor blob by sha) is forge's; not edited here.
Still open and named: 25 FAIL rows are per-lane findings; 51 PASS* rows read an absolute host root (Q62's class, 48 before this tick); path-claims is forge's two-character citation at SMELL_CHECK_MONOTONIC_REFACTOR_VERIFICATION.md:95-96, so tools/verify.py is 8 legs / 7 pass / 1 fault both before and after; promoting volt CHECK to a tools/verify.py leg stays the operator's call. Index re-rendered in this commit.
Record at wizard-work/volt/q65-2026-10-24.md, queue addendum appended, index re-rendered in the same commit (11096 lines, 52 ids unchanged -- nothing here opens a record).
The item narrows rather than closes: 88 committed verifiers = 71 graded by q65_runner + 4 invoked by a committed line + 8 declared one-off + 5 executed with no rc (TIMEOUT). The five are named, and a per-file window is what grades them.
The census had no rev, so its classes read as current forever -- the archive rule it exists to enforce. emit_baseline now writes the short sha into the header.
Re-emitted after the four graded rows landed: gate-run 71, wired-invoke 4, declared-one-off 8, undeclared 5 (88 rows, 88 committed verifiers at 2843d50). The 5 undeclared are the TIMEOUT rows: executed, no rc, so not graded and not tolerated as clean.
Four files the prose fix released from the wired class are graded by execution, not by reading: forge/verify_roundtrip_refactor.py PASS 0, ledger/verify_arxiv_csv.py PASS 0, ledger/verify_github_python_repos.py FAIL 1, sphinx/archive-remember-r1-figures verify_archive.py TIMEOUT 120s. Rows: PASS 53, FAIL 18, TIMEOUT 5; the 5 ungraded carry no rc and stay unwired in the census.
wizard-work/volt/CHECK is the caller the runner did not have -- the Q65 defect one level up. It runs the selftest, --check and the scan; none of them executes a verifier (regrading is --emit, minutes). Promoting it to a tools/verify.py leg is a decision for the operator, not for a lane: recorded here, not assumed.
unclassed() asks the tree at --rev, not the ledger, which verifier is named by nothing. At b01ec0e that is 8 UNCLASSED of 88 and --check returns rc 1 where it previously printed "0 fault(s)": the population was whatever the class ledger listed, so the instrument inherited that ledger omissions.
Also: emit() wrote "rev: HEAD". A ledger whose law is "the rc at a named rev" cannot name a moving ref -- it is pinned to the short sha now. And the single timeout header figure was the LAST --merge window, standing over rows run under 45s/90s/600s windows; the window is per row, named in the TIMEOUT note.
Selftest 9 -> 11 cells: a committed verifier in neither ledger is UNCLASSED, and a class row covers it.
gate-run 68, wired-invoke 4, declared-one-off 8, undeclared 8. The arithmetic is the point: the row count equals the committed verifier count at the rev, so a verifier that appears in no row is now a measurable difference rather than a silence.
Two defects, both measured at b01ec0e before the change.
1. emit_baseline() skipped every WIRED row, so the ledger could not express a file it had never classed. Eight of the 88 committed verify_*.py sat in no class and no ledger while both instruments printed clean. It now writes a census: one row per committed verifier, four classes (gate-run / wired-invoke / declared-one-off / undeclared), third column carrying the evidence the class rests on.
2. wiring_refs() tested INVOKE_PATTERNS against the whole git-grep record ('<path>:<line>:<text>'), so an interpreter token could come from the REFERRER'S OWN FILENAME -- 'extract_github_python_repos.py' supplied 'python' and a prose mention supplied the basename. The shape is now tested on the line TEXT only. And a shape inside unexecuted text is not a shape: py_line_kinds() marks docstring/comment rows dead via ast+tokenize, and a string-only line counts only when the line also carries an exec-shaped call (subprocess/runpy/executable...).
RED at HEAD, the three prose lines, from the committed scanner (git show HEAD:...): WIRED ledger/verify_arxiv_csv.py, WIRED ledger/verify_github_python_repos.py, WIRED sphinx/archive-remember-r1-figures-2026-10-23/verify_archive.py. GREEN after: all three are mentions, and the four genuine invocations are unchanged (spark/run_verify.sh, lc-coverage and lc-mm-dossier mutation_battery.py, transformer CHECK.py). Selftest 6 -> 10 cells, the four new ones built from these exact shapes.
path_claim_check graded it NEW (not in the baseline) and tools/verify.py was 8 legs / 7 pass / 1 fault on it. The file is the lane's own evidence for a closure already on the record; the remedy named in the Q63 addendum (reply 91 §1) was this commit and no content change. Blob unchanged: 10108 bytes, mtime 18:34 as written.
3aa745c appended the record and ran queue_index.py without --write, so
index-current read STALE at line 63 and append-audit named 3aa745c
index:no. The re-render alone fixed index-current but not append-audit,
which grades the newest queue-touching commit in its own tree; this
append supersedes it with the render in the same commit. The record's
measurements are unchanged: reply_check 333 files faults 0, path-claims
17 missing 0 new.
My previous commit appended to WORK-QUEUE.md and re-ran queue_index.py
without --write, so the index printed its summary and wrote nothing;
verify.py's index-current leg caught it: "STALE: rendered document differs
at line 63", and append-audit named 3aa745c as the stale commit. This is
the same tree after the one-flag fix -- 11 insertions, 11 deletions, no
hand edit.
Appends the record at the foot: the red (4 DUPLICATE_NUMBER), the measured
next free integers (67 / 12 -- reply 87's 59 was taken), the identity
repair (numstat 1/1 each), the empty-by-measurement reply ledger, and the
two RENAMED rows the rename costs. path_claim_check --fail-on-missing ->
17 missing, 0 new. Index re-rendered in this commit.
reply_check.py read 4 DUPLICATE_NUMBER faults at 09dfa0f, two pairs:
the 10-15 merlin file created by 5d00265's rename collided with the
10-14 merlin-58, and 2026-10-24-to-mercury-11 collided with 10-22's.
Numbers are unique per recipient series, so the later file in each pair
moved to that series' measured next free integer (merlin -> 67, not the
59 reply 87 proposed: 59-66 are all taken). Filename and the heading's
own number token only; every body line byte-identical. reply_baseline's
one row went STALE when 5d00265 renamed its file, and no file carries
HEADER_RECIPIENT now, so the ledger is retired empty.
reply_check.py --repo . -> 333 files, DUPLICATE_NUMBER 0, faults 0, rc 0 bare.
The module's docstring claims 'results match the C reference' and no cell
graded it: _MULT ...1757->...1751 and the md5 hi^lo fold -> hi both passed
7/7 while remapping 99.0% and 99.0% of 10k string keys at n=100 (measured,
transcripts in volt's 2026-10-24-jump-golden artifact).
- jump_ref.c: the Lamping & Veach figure-5 reference (gcc -O2), committed
beside the test so the vectors can be regenerated from the repo alone.
Measured against the Python module on 630,910 (key, n) pairs: 0 mismatch.
- test_golden_vectors: 0/5->0, 2^64-1/100->92, _to_u64('key-0') and
jump_hash('key-0',100)->40, jump_hash('order-42',1000)->667. The string
cells are the only ones that can see the fold: m2_fold moves 0% of
integer keys by construction.
- test_minimal_movement: lower bound at 0.5x expected, so a sticky hash is
graded, not merely absent from the failures (measured ratios 0.974-1.047).
- test_uniformity_other_bucket_count: n=97 prime, +/-10%, observed -5.28/+4.57.
Suite 7 -> 9 tests, both green unmutated; control and three mutants measured
before/after in the artifact.
The ledger attach loop already visits every MISSING row, and the old return
statement then re-walked missing.values() three more times to recover matched,
new and tolerated. Those three tallies are now accumulated in the attach pass
itself: known and new are exact complements within one row, and matched /
tolerated are read from the record just written. Four traversals of the row
set become one.
Output is unchanged, and that is measured rather than asserted. Against
54961d4's copy of the same tool, twelve invocations over the live worktree are
byte-identical in stdout, stderr, exit code and --json payload: default,
--json, --only missing|docrel|external, --context, --fail-on-missing, an empty
ledger (the `new` list, all 14 counted), the same graded (rc 1, not 0 and not
2), a ledger row naming a class outside the closed set (rc 2), a truncated
ledger (matched/tolerated/dead at once), and a non-HEAD --rev. Selftest is
31/31 on both copies.
No speed claim: five timed runs put HEAD at 6.94/6.92s median and this at
6.70/6.90s -- inside the noise, because the 1598-path filesystem scan dominates
the row walk. The change buys passes, not milliseconds.
volt
volt
volt
volt
Volt