D0026 — Clinical work, not scaffold: the refocused ten and the v1.0 question

[obotclaw[bot]]

The full argument is on the artifact page: https://jwildfire.github.io/obot.roadmap/reports/decisions/2026-08-21-clinical-priorities/

The question you asked. Is the framework good enough to push towards open.gismo v1.0? Yes for building clinical work, no for calling anything 1.0 — and only the first is about the framework. In the last month the machine delivered a kidney-safety explorer, a Kaplan–Meier survival chart whose estimator agrees with R's survival package to ten decimal places, three participant-level safety metrics and a study census, each through the same lane. What is missing is bookkeeping a version number rests on: nobody has written what 1.0 contains, the design question gating every app requirement has sat open since 12 July, and the demo study has failed its weekly rebuild on 3, 10 and 17 August, with the issue filed after the first failure never ranked since. Recommendation: start the push now with clinical build work, and treat those three as the first hour rather than as a phase.

What moved under the gap analysis. It is four weeks old and seven of its statements have changed. Two charts shipped since — the kidney explorer and Kaplan–Meier, the second of which was on its headline list of things four or more competing platforms ship and we did not. Its "zero review-workflow capabilities filed" is out of date: you directed the action-log requirement three days after it was written and it covers three of the ten. Its "nothing flags a participant" is half wrong now. And one thing got worse — the requirement that would have carried one shared filter and one shared selection across charts closed on 15 August with the mechanism unbuilt, so that capability has no issue anywhere.

The five questions.

• V1 (D0026.1) — Is the framework good enough to push towards 1.0, and does the push start this weekend? Start now with clinical work / harden first and push in September / define 1.0 first. Recommend: start now, bookkeeping alongside.
• V2 (D0026.2) — The refocused ten: accept the proposed order or change it. Recommend: accept, and keep the board-write failure and commits-under-your-name visible on the bench.
• V3 (D0026.3) — Which single piece of clinical work is the weekend experiment? Recommend: the R widget catch-up, which fixes a liver chart the R package is drawing wrongly today and runs the whole lane end to end.
• V4 (D0026.4) — Which missing charts get filed now? Recommend: file risk-difference screening, mean change with intervals, and the context set as chart requirements; file one-filter-one-selection as a platform requirement; hold per-participant change-since-last-review inside the review-layer design.
• V5 (D0026.5) — Does the review layer become 1.0 scope, or is it deferred in writing? Recommend: a design pass this quarter, and out of 1.0 in writing.

The proposed ten, in order. The R package's wrong liver chart (#164) · the census rebuild you approved (#274) · the data-coverage chart (#291) · the demo study that has not rebuilt (#153) · settle what GitHub is to the app (#34) · take Experimental off the survival chart (#182) · R widgets for the two newest charts (#165) · the review-layer design pass (#139) · mine the old renderer trackers (#33) · decide what the survey produces (#176). The migration is off by your word and is not ranked.

Ranking is obot-prime's call; this is a proposal it will act on. Nothing was filed, ranked, closed or built.

---

Drafted by Claude Code using Opus 5, background worker W0109.