Everything §0.3 built — the deck's emitter (EI.2.02), its command listener (EI.2.03) and the portal's relay (EI.2.04) — rebuilt from source and run.
- Build commit:
db206b90dddf5ce785dd93f5b765d058371dc291 - Date: 2026-09-19
- Browser: bundled Chromium via
CHROME_CHANNEL=chromium, per CLAUDE.md's proportionate-verification decision. - Supervisor: every browser harness ran through
tools/presentations/run-supervised.py, which refuses to start below 3 GiB available and stops the process group below 2 GiB. It reported 13.1–13.7 GiB available throughout.
What ran, and what it said#
| Report | Command | Result |
|---|---|---|
build.txt |
build-eve-oshun.py |
Built 1721 slides, 728 sources |
check.txt |
build-eve-oshun.py --check |
Checked 1721 slides, 728 sources, exit 0, no stale artifact |
python-tests.txt |
unittest discover -s tools/presentations/tests |
558 run, 557 pass, 1 pre-existing failure |
node-tests.txt |
node --test tools/presentations/tests/*.test.mjs |
521 run, 519 pass, 2 pre-existing failures |
harness-deck.txt |
eve-oshun-deck.mjs --guide agentic-studio |
exit 0, including PASS messages to a framing host over HTTP, and silence under file:// (EI.2.02) |
harness-center.txt |
presentation-center.mjs |
exit 1 — 11 checks pass, then a pre-existing colour-contrast audit |
harness-portal-relay.txt |
testPortalRelay standalone |
PASS — the centre harness stops before reaching it |
The build was stale, and --check is what says so#
Running --check against the committed index.html before this rebuild
printed Stale generated artifact: index.html and exited 1.
The cause is worth writing down: the commit hook runs prettier over
center.js, and it did so after the build that inlined it — so the
committed index.html carried an older formatting of a script the source no
longer had. A build followed by a hook that reformats a build INPUT leaves the
output stale, and nothing but --check notices. This directory's build fixes
it.
The three failures, each measured as pre-existing#
None was introduced by §0.3, and each was checked rather than assumed.
test_the_seeded_crosswalk_holds_against_the_tracker_as_it_stands(python). The same gate recorded during the Eve Task Board cut-over rehearsal on this date: its output was byte-identical (diff -qclean) between a scratch clone and the untouched worktree, and it is aboutT.20.08in the isis-chroma-runpod tracker having shipped without an ungated slide teaching it. Nothing gates oncheck-generation-crosswalk.py: it is referenced by no workflow, hook or package script, and was added 2026-09-18.pinned-tara-ritual-assembly.test.mjs(node, 2 assertions). It trips overapps/oshun/mobile/dist/_expo/static/js/*.hbc— build output that is gitignored and dated 29 April, months before this work.- colour-contrast on
li[data-node="realize"] > .step-actor(presentation-center.mjs), which resolvescolor: var(--actor)atdeck.css:4329. Measured pre-existing two ways:deck.cssis byte-identical to this session's starting commit (sha256323d7edeea57aa31both), and the flagged markup appears in exactly one deck —portfolio-map.html— at session start and now. The audit's inputs are unchanged, so its verdict is. Changing--actoris a design-token decision across every journey step in the library.
Because that third one stops presentation-center.mjs at check 12, the portal
relay check registered after it never runs there. It was run standalone under
the same supervisor instead, and its report is harness-portal-relay.txt.