Scope: the fourth chapter of the Use Eve track, rewritten in the Eve edition — two manuscripts merged into one, eighteen slides re-treated and re-voiced, one content slide added, six chrome slides added, and every binding re-reviewed. The brief records what the chapter is for and why each treatment was chosen; state.json holds the exact counts, durations and tool results.
What was checked#
- Build freshness.
build-eve-oshun.py --checkpasses on the delivered tree: 1,378 unique slides, 658 sources, 75 guides. The chapter is 25 slides and 33.1 minutes of narration. - Teaching crosswalk.
check-eve-capability-teaching.pypasses across 117 slides. Seventeen assignments for this guide carry the rewritten titles, and every one of their boundaries was re-read against the new slide and still holds, so none needed rewriting. - Narration. All 25 tracks rendered and verified:
--check --idsover the whole chapter exits 0, which checks the fingerprint, byte count, SHA-256, probed duration and a full decode against the authored pause plan. The other 1,353 tracks are untouched. - Browser rendering.
eve-edition-review.mjs --guide eve-delivery-operatorsweeps all 25 slides at 1440×900, 1280×720 with audio chrome, 390×844 and in print with no findings. Fifteen slides were also read at 1440 by eye, including every new layout the chapter introduces. - PDF. Exported from the same sweep and held to the exact-text contract by
check-eve-specimen-pdf.py. 25 pages pass. - Suites and lint. Python 449 passed with 2,729 subtests; Node 443 of 445; Ruff, ESLint and Prettier clean on everything changed.
What was fixed on the way#
- A banner-shaped capture lost its legend at 1280. The short-screen rule that moves a stack legend beside its capture assumed the capture would be narrower than the row. A capture seven times wider than it is tall takes the whole row and left the legend spilling 80px off screen. The renderer now marks a capture whose width is at least three times its height, and that layout keeps its legend underneath.
- The spill check could not tell a scroller from a defect. The four-queue table is meant to scroll horizontally on a narrow screen, and the harness reported its off-screen columns as lost text. It now exempts text inside a horizontal scroller the reader can actually reach — and the exemption was verified not to weaken the check, by injecting an unreachable overflow and confirming the harness still reported 30 findings.
What was not checked, and what was not done#
- The two new captures the plan asked for were not taken. The fleet
attention view and the assistant-health view are admin surfaces with no
component under
apps/oshun/web/src/components/assistant/, which is the only tree the capture harness mounts. Separately, a capture taken on this host renders under Chromium 148 where every committed asset was made under 152, which would make the recorded capture environment false for the new images. Those two slides are a card grid and a compare panel instead, and the chapter takes its real capture fromwork-delivery.png, already provenanced. The capture item stays open in the TODOS with this reason. - No human reader session and no listening review at normal speed.
- The narration was rendered on this host's CPU rather than the Apple GPU the rest of the library used; only the MP3 bytes differ.
test_v2_rollback_sessionstill fails. It guards prior teaching against a published baseline, and each delivered chapter rewrite trips it. Re-baselining that guard is a decision for the user, so it was left alone.