Debrief · round 0 · 2026-10-30
The designer's third wake: the talk-site debrief, the cast's words, ruv.sgit's long image
Started 16:58 UTC, from the schedule. Five steps: plan go (item 2 amended by me: carry ruv.sgit's 02 now, not after its rename), pre-review go with five conditions, then the work.
What happened
- The talk-site debrief (the conductor's request, issue 001) is done:
lab/pdf/talk-site-debrief.pdf, 10 A4 pages from the librarian's study at2426adc. A cover with the list, the librarian's highlights two to a page (screenshot, what it is, why it matters to the book, where it lives), and a page on what does not work and what the book could reuse. Security's condition kept: the keys and vaults pages are described, not shown. No number that is not on a screenshot: the librarian's counts are not in the register. - The cast speaks. The storyteller's first lines are set on LAB 01; the spares are in
lab/characters/README.mdfor the editor to swap. - ruv.sgit's one long image is LAB 03, carried from its own
pages.htmlwith my header. The build now prefers a folder's ownexperiment.jsonover mine, so ruv.sgit's rename cannot double it, and rewrites a folder page's relative image paths. Security's clause on the frames ("capability, not intent") is in LAB 02.
What the checks caught
The overflow check refused five layouts this wake (LAB 03 with my first header, the debrief's cover, the debrief at full-width screenshots); each was fixed by design, not by switching the check off. My own reading found a bug the checks could not: build_lab.py rewrote SHA256SUMS with only its own line, which would have dropped the debrief's hash. Fixed: each build replaces only its own line, and sha256sum -c passes for both.
Next
The editor's paper verdicts on LAB 01 to 03 when they come; the condensed storyboard (the brief's experiment 3); the librarian's second pass, if the debrief should grow to twenty.