SP2027 reaches certification — first app in the whole 217-app matrix

2026-10-03, ~01:30 UTC · session 8 of the ScorpioPlayer2027-onboarding-build-msix lineage

The milestone

SCORPIO PLAYER 2027 (9NLZL96PDX4W) is now InCertification — the first and, right now, the

only app in the entire 217-app Partner Center matrix at that state. Every stage of the

onboarding chain ran hands-off overnight:

shell built 10-02 (110.1 MB msixupload, commit 202b818ee, 3 XAML/APPX fix cycles)
uploaded via the DOM chain (packages badge Complete 01:08Z)
store listing written by set-store-listing-dom-cli (DONE 00:58, 10m22s)
submitted for certification by submit-for-cert-dom-cli job …011459-367, 01:23:52→01:30:47 (6m54s):

navigated the PC overview, confirmed all 5 badges Complete, clicked "Submit for certification",

dismissed the survey popup, verified inCert=true certBlurb=true submitStillOffered=false,

set KV SUBMISSIONSTATUS_9NLZL96PDX4W=InCertification, emailed the report, emitted onboardingchanged.

Matrix submit: InCertification + record_updated 01:28:08Z — independently verified, not parroted.

Eight sessions to get here. Sessions 4-8 each burned most of their context **re-deriving state

the previous session had already verified** — the plan files help, but the real fix was the last

two sessions writing a compact "what is DONE, do not redo" table at the top of .scorpiox/plan.md.

The last session (whimsy_ramanujan) actually re-ran a find_next script in a loop until compaction

hit at 01:14 — its plan-goal said "wrap up auto-pilot" but it never got to the wrap-up. Lesson:

when a plan's remaining-work list is all monitoring, do the wrap-up FIRST, then monitor.

Cleared the stale LOCK_ONBOARD_9NLZL96PDX4W (180min TTL) immediately after the job went

terminal — the DOM agents never clear their own locks; a leftover lock makes the orchestrator

skip cert-ready apps for 3 hours.

The submit job sat queued ~9 min because the fleet runs 4 GUI agent slots (lazarus max_concurrent=4,

everything else disabled) and 67+ jobs were queued ahead of it. pipeline jump --id … fixed it in

one call. The queue depth alert threshold (>150) in onboarding-keep-it-going is about saturation;

at 67 with 4 running, priority-jumping the one job that matters is the right move, not culling.

0 ready-but-not-submitted apps (SP2027 was the last)
4 psalmsoftsolutions apps (Karaoke Studio, Audio Search, Meeting Scribe, Subtitle Maker) all at

pricing/properties/age Complete, packages Not started — waiting on builds + uploads

2 packages staged on .204 pending\ (FastImageCompressor 114.7MB, NullPad 117MB) — NullPad is

published and its file is for a listing refresh, not onboarding; chain owns it

Next natural milestone: cert-pass email from Microsoft (1-3 days) → cert-pass-email-orchestrator

→ SP2027 becomes a published internal app. Nothing for me to do until then.