Skip to report
Chief / BettterAI Design Adapter13 November 2026

Her 12 Nov review in blazers v1.6.23 and trousers v0.11 - live on prod and on the review site

Every item she sent on 12 November and her Slack note on ALVA's dart now carry what was built; both engines went live on prod on the owner's instruction ("deploy latest", "and merge"), with the failed release checks kept as measured.

Live on prod: blazers v1.6.23 (11:31 UTC), trousers v0.11 (09:01 UTC); on adapter.bettter.us for her review

Grade A and B blazer pairs that match the pattern maker, per version

v1.6.1889 of 104 pairs
v1.6.1985 of 104 pairs
v1.6.2086 of 104 pairs
v1.6.2187 of 104 pairs
v1.6.2289 of 104 pairs
v1.6.2394 of 104 pairs
Source: adapter.bettter.us Versions page data (site build of 13 Nov, every version re-tested on all 141 pairs)

ALVA bz-105 before: v1.6.22

The bottom of the ALVA front of bz-105 in v1.6.22: the master's dart (orange) lies on the raw material's dart (green), but the step of the cut line stays 17 mm to its right.
v1.6.22 (export bt_v1622b): the dart moved onto the old dart, the step at its foot did not. Black: sew line, red: cut line and notches, green dashed: the raw material.

ALVA bz-105 after: v1.6.23 (D36)

The same front in v1.6.23: the step's edge now continues the dart's line; the hem's middle and dart-side notch pairs moved with it, the front edge's pair stayed.
v1.6.23 (export bt_v1623): the step moves with the dart (20 mm here) and 4 of the hem's 6 notches move with it, as she drew in her 12 Nov message. Only the 7 ALVA masters have this step.

Release checks against live v1.6.22 (agreement with her verdicts)

Prod library v1.6.22 (of 111)105 pairs
Prod library v1.6.23 (of 111)106 pairs
Dataset v1.6.22133 pairs
Dataset v1.6.23133 pairs
Source: docs/iterate/gate.py: prod library data on the same 112 pairs (8 on raw material 0171-B 56 left out), her dataset 141 pairs

How it went (13 Nov, UTC)

  1. Streams

    Sleeves (S2), body (B2), trousers (T2) built her 12 Nov items; merged 07:16

  2. ALVA dart

    Her Slack note relayed by the owner: the step follows the dart (D36), 08:05

  3. Merge

    PR #76 08:58; trousers v0.11 live 09:01

  4. Full run

    141 pairs + 120 library pairs, gates, notes, UA/PT texts

  5. Deploy

    PR #77 11:25; review site; blazers v1.6.23 live 11:31; live job passed

What changed for her
  • Sleeve vents come out alike in width, length and slope, made alike at the sleeve hem, never by changing the cap (bz-082, bz-214 keeps both old vents).
  • LORCA and FARROW top sleeves get no vent, as the master; the back vent's two halves are made alike where nothing gets worse (bz-124 asked).
  • Sleeve hems are straight between their corners (bz-117, bz-139).
  • A raw material more than 5 cm longer keeps the master's length (bz-066, bz-114, bz-117, bz-118).
  • The back moves a little for the master's step, and the side panel's armhole corner moves in for it, as she answered on bz-022 (bz-146 adaptable).
  • The pocket notch moves onto the old pocket (bz-092); the old front corner is kept (bz-150); ALVA's dart sits on the old dart and its step moves with it (bz-103, bz-105).
  • Trousers: the back waist may curve up to 5 mm, the length is the raw material's (no NORMAN limit), the fly front may come up to 3 mm out where needed.
Not achieved and accepted with the owner's instruction
  • Release checks failed and were accepted as measured: bz-068, bz-114, bz-120 (prod library) and bz-067 (dataset) now adapt with the master's length where her verdicts, given before her 12 Nov length rule, say not - asked again (rq-length-longer-2); bz-108 adapts on the dataset where she said not (asked).
  • Her fixed-note checks: 42 back (v1.6.22: 43); 5 came back - the master's steps on bz-016 (side panel), bz-105 / bz-134 (top sleeve), bz-138 / bz-140 (under sleeve). Next version.
  • Trousers v0.11: tr-007 adaptable as her 12 Nov verdict, where the tracker (the trousers check) says not.
  • Live trouser jobs are still refused by the library (their SDAs are pending validation); the same job passed on lab v0.11.
  • Library raw material 0171-B 56 has no readable front since 11 Nov: 8 pairs error on any version.
  • QA live stays v1.6.14 / v0.4 (lab v1.6.23 / v0.11). No native-speaker review of the new UA / PT texts (224 each, machine translated, checked for placeholders).
Questions for her (4 blazers, 3 trousers)
  • bz-118: shorten by 0.4 cm, or keep the master's length?
  • SCOOP NECK: how far from the neckline must the chest pocket be?
  • Over 5 cm longer, the master's length kept: are bz-067, bz-068, bz-114, bz-120 adaptable now?
  • bz-124: the old back vent's top is slanted, the side panel's square - how to make them alike?
  • Trousers: tr-063's fly front up to 3 mm, tr-065's waist moved less, shortening more than the raw material needs.
Verification
  • Full test suite before PR #76: 953 passed, 5 failed (2 fixed, 3 explained on the PR); targeted suites pass on the final branch.
  • Relevance gate: blazers 12 shown / 0 stale; trousers 4 shown / 0 stale.
  • Published site: v1.6.23 and v0.11 current, 62 of 62 and 14 of 14 of her new notes shown; 401 without the site login.
  • Lab jobs passed (blazers bz-058, trousers tr-005); live job on blazers v1.6.23 passed after the promotion.
  • Rollback targets: v1.6.22 / v0.10 (engine), a copy of the review site from before the publish.