troupe-standup/recoveries/helm-adoption-sequence-readiness.md

Helm adoption sequence — measured readiness (Q45S11 / Run 31)

Run 31, Helm lane. Scratch tree /root/crucible/q60/; instrument measure_r31.py, artifact out/R31-ADOPTION-READINESS.json (host-local, E13). Clock at the run: 2026-10-14T20:47:46Z (the census) / 20:49:44Z (the re-run). Nothing deployed: no cp into /root/.helm/bin/, no --deploy, no write to the live document root. decisions/ and inbox/ untouched, no another-party file edited, no secret printed.

1 What was measured, and why

The Helm lane's executable set has been empty since Run 30: Q9, Q13, Q19 and the career-lane Q53 are Alexander's, Q42 is Zephyrus's, the Mercury Q52 is closed and the Mercury Q53 (state.db / WAL) is Mercury's, and the queue's genuinely open archive items are other lanes' or gated. The one thing that has never been measured as a set is the adoption sequence the pipe has carried since Run 29 and re-confirmed at Run 30 — three files, in one order, whose payload is the thing Alexander would have to execute. Every file has been verified individually (by this lane and independently by the pipe at c3b464f/054aa43); the bundle — the three files together, the staging, the green pair, the re-render payload, and the live delta around all of it — has not. That is this run's unit.

The sequence, as carried: adopt renderer 306d4bee… (22,731 B) first, then adopt collector 4e45021d… + gate 125754b270bf… (161,122 B / 3,053 lines / 135 registrations), then re-collect → re-render → deploy. Renderer first because a re-render on the deployed renderer publishes the internally inconsistent page G11 measures.

2 The three candidates, each against its stated figures

Measured on disk this run (measure_r31.py §"candidates"), not read from any prior record:

file	path	bytes	sha256	lines	stated match
renderer	/root/crucible/q59/helm-render-g11.py	22,731	306d4befee7c9874edc6a7467945420e94aa6a0f0445e1955641f07bc046c8a4	512	bytes ✓ sha ✓ lines ✓
collector	/root/crucible/q48/helm-collect.py	29,954	4e45021d80395d3f03fc58d7e99610e3c1c0d2fcf9c85f1db50b16cfd2043c6c	720	bytes ✓ sha ✓ lines ✓
gate	/root/crucible/q59/helm-selftest-q59.py	161,122	125754b270bf6b0f55911f407fa9c36c6161b45a71e646fa36a70db096d7de93	3,053	bytes ✓ sha ✓ lines ✓

All three end with a trailing newline. All nine cells true. The adoption digest triple is sound as a set.

One defect of mine, logged first: the first run's STATED dict carried the collector digest with one hex character wrong at index 15 (…80395d3f**3**fc… printed as …80395d3f**f**fc…), so the instrument reported sha_matches_stated: false for the one file whose digest it transcribed rather than measured. Caught by the instrument itself (the other two, transcribed identically, matched) and corrected before the run was recorded. The measured value is 4e45021d80395d3f03fc58…; the lane's own earlier record and the pipe's reply 56 both carry it correctly, so nothing published is affected — only this instrument's first output was. Class: a digest quoted beside a path must come from a read of that path, never from memory (the rule this lane logged after the 4e3cca9c/decd3350 pair).

3 Staging, and the green pair

The three candidates were copied into /root/crucible/q60/adoption-bundle/ with shutil.copy2; each staged copy re-hashed byte-identical to its source (sha_identical: true ×3, at 22,731 / 29,954 / 161,122 B). The staged collector was then handed to the gate by name:

HELM_COLLECTOR=/root/crucible/q60/adoption-bundle/helm-collect.py \
  python3 /root/crucible/q60/adoption-bundle/helm-selftest-q59.py

rc 1, helm-selftest: 184 passed, 1 failed, 15,841 B stdout / 2,090 B stderr, 15.15 s. The one failure is G10's deployed-page row — the gate asserting the live page carries the qualifier it measurably does not, which is exactly the red row that makes the re-render necessary. This reproduces the 184/1 matrix the G11 record carried and the pipe re-ran at c3b464f, on the staged copy rather than the scratch original.

4 The re-render payload

The real snapshot through the fixed renderer's own CLI, at this run's clock, to scratch:

python3 /root/crucible/q60/adoption-bundle/helm-render-g11.py \
  --out /root/crucible/q60/expected-deployed.html --as-of 2026-10-14T20:47:46Z

rc 0, 11,811 B / sha256 f47262f8f7c00f1fda0091547273388411b2f3d312f87030724d625c53a9aa60, report render clock 2026-10-14T20:47:46+00:00 · snapshot age 6d 3h 34m / verdict at render: 0 green / 6 not green (5 stale, 1 unreachable). The payload carries as of true and the match (unverified) span true — the two things the deployed page lacks. The re-render is +1,946 B over the live page (9,865 B) and its digest differs (27ea2ad2… was the first run's per-clock value; this run's is f47262f8f7c0… — the age string is relative, so the bytes are a per-run value and the digest must never be cited as the page's; the durable claim is the two content flags and the delta class).

5 Live state, delta 0 around everything

Census (bytes + sha256 + mtime_ns + ctime_ns) taken before the gate, after the gate, and after the render; and /root/.helm/bin/ globbed for every entry (five files plus the __pycache__/ an adoption must not copy):

object	bytes	sha256	mtime_ns	delta
deployed page /var/www/alexander-whitestone/helm/index.html	9,865	427b7b54c69b18d2…	1791709034017837233	✓
snapshot /root/.helm/data/snapshot.json	14,039	aa842b0fd88037f8…	1791479579475780303	✓
held pin …/held/stackchain-home.html.sha256	120	5b6d70716153c3f4…	1791478687099553071	✓
helm-collect.py	27,907	ee0b36a82f23fcf3…	1791709575259049023	✓
helm-render.py	21,766	5c67473d89283c3d…	1791717163756240575	✓
helm-selftest.py	33,647	7ed1bca8f9da624f…	1791717211359222589	✓
helm-validate.py	3,556	12570786ba70…	—	✓
summarize.py	1,463	738699986778…	—	✓
__pycache__/	4,096	(dir)	—	✓

Delta is true on all rows across both the gate and the render legs, and grep -c HELM_COLLECTOR /root/.helm/bin/helm-selftest.py = 0 — so Q40 stands and nothing of Q31–Q45S11 is in production. Sandboxes 0.

6 What this does and does not establish

Established (Documented, this host, this run): the adoption bundle's three files match their stated figures as a set; they stage byte-identically; the gate is green-but-one against the staged collector, the one being the deployed-page row the re-render exists to clear; the re-render payload carries both missing markers; and nothing live moved around any of it.

Not established, and deliberately not run: the 161 KB gate's own 184/1 is a recorded run reproduced here on the staged copy (standing rule — the gate is not re-run gratuitously); the post-adoption re-collect's output is unmeasured because the deployed collector cannot run without itself being the accident Q34/Q38 guard against; and the render payload's bytes are per-run (the age string), so only its content properties are cited.

Limits: the record cites no out/ path as forge-verifiable (E13 — every out/*.json is host-local, 0 paths under out/ in the repo tree). The staged bundle lives only in /root/crucible/q60/adoption-bundle/. The __pycache__/ row is a directory census entry, not a digest.

Nothing outstanding to the pipe; no new queue item opened. The Helm executable set is empty apart from the adoption sequence, which is Alexander's. Standing human gates unchanged: the §4 sequence (adopt renderer 306d4bee… → adopt collector 4e45021d… + gate 125754b270bf… → re-collect → re-render → deploy), the stale manifest_match pin, Q13's publication call, Q9's checksum handover, Q19's onboarding.

— Merlin (Run 31)

7 Corrections, Run 32 — reply 57's C1, C2 and the index nit

Appended, not edited: §4's text above stands exactly as written and is superseded by this block, per the lane's append-only rule for records. Instrument /root/crucible/q61/measure_r32.py, artifact out/R32-REPLY57.json (host-local, E13), run on this host at 2026-10-14T23:00:28Z. Nothing deployed: no cp into /root/.helm/bin/, no --deploy, no write to the live document root.

C1 upheld — the digest was quoted beside the wrong clock, and the parenthetical was wrong in a second way. The real snapshot was re-rendered through the verified renderer (306d4befee7c…, 22,731 B / 512 lines, sha_matches_stated: true measured this run) at three clocks, all into this run's own scratch tree:

--as-of	bytes	sha256	render clock / snapshot age	as of	match (unverified)	vs live page
2026-10-14T20:47:46Z	11,811	27ea2ad2ccf03f18…	20:47:46+00:00 / 6d 3h 34m	true	true	+1,946 B
2026-10-14T20:49:44Z	11,811	f47262f8f7c00f1f…	20:49:44+00:00 / 6d 3h 36m	true	true	+1,946 B
2026-10-14T23:00:28Z	11,811	7dbf6c5f3586ec06…	23:00:28+00:00 / 6d 5h 47m	true	true	+1,946 B

The corrected pairing: §4's code block should read --as-of 2026-10-14T20:49:44Z, and f47262f8f7c00f1fda0091547273388411b2f3d312f87030724d625c53a9aa60 is that clock's digest. Three independent objects agree, all read this run rather than remembered: the R31 artifact's own clock field is 2026-10-14T20:49:44Z; its render report line reads render clock 2026-10-14T20:49:44+00:00 · snapshot age 6d 3h 36m; and the on-disk /root/crucible/q60/expected-deployed.html is 11,811 B / f47262f8f7c00f1f… (mtime_ns 1792010999833643759). The 20:47:46Z in the code block was the census clock from line 5, transcribed into the render's --as-of; that clock produces 27ea2ad2ccf03f18….

The second defect is mine and the pipe did not name it. §4's parenthetical calls 27ea2ad2… "the first run's per-clock value". There was no first run — this record's run made exactly one render, at its own clock. 27ea2ad2… is the digest of a render at the census clock, and Run 32 is the first time any run produced it.

Nothing in the finding moves. All three clocks give 11,811 B, as of true, match (unverified) true, +1,946 B over the live 9,865 B page, and a digest different from it — the two content markers and the delta class hold identically, which is exactly the durable claim §4 and §6 already scoped the record to. The defect was the pairing alone, exactly as the pipe graded it.

The index nit, upheld and made precise. §2's self-description reads "one hex character wrong at index 15". Computed here: the first differing index is 16, and the difference is an omitted 0, not a substitution — the true segment is 80395d3f03fc (indices 8–19) against the printed 80395d3fff; index 15 is f in both. The substance stands and is confirmed against the file: measured_from_file: true for 4e45021d80395d3f03fc58d7e99610e3c1c0d2fcf9c85f1db50b16cfd2043c6c against /root/crucible/q48/helm-collect.py.

C2 upheld and filled in place. git -C /root/troupe-standup diff --numstat 6314927 7bcd373 = 51/0 on WORK-QUEUE.md, 92/0 on the check-in, 144/0 on this record = 287 insertions, 0 deletions (--shortstat → 3 files changed, 287 insertions(+)), reproducing the pipe's figure to the row. The literal <parent> <this> template at checkins/2026-10-14-merlin.md:1458 (line 1459 at 7bcd373, one line higher before the 48f0e65 landing fill) now carries the measured revisions and split, filled in place at wc -l 1,476 → 1,476, so no line citation below it shifts.

Live state, delta 0 across the whole run (bytes + sha256 + mtime_ns + ctime_ns, taken before the first render and after the last): page 9,865 B / 427b7b54c69b18d2…; snapshot 14,039 B / aa842b0fd88037f8…; held pin 120 B / 5b6d70716153c3f4…; all six entries of /root/.helm/bin/ including the __pycache__/. grep -c HELM_COLLECTOR /root/.helm/bin/helm-selftest.py = 0 — Q40 stands, so nothing of Q31–Q45S11 is in production. Sandboxes 0.

— Merlin (Run 32)