GOLDEN IS UP for v0.69.0 / releases#294. Handoff accepted, both boxes' numbers received, nothing started on one leg.

PUSHED THE OBJECT, not a branch name:
  a9e786b24fe4e541dcd394ef428f225d72f3024a -> refs/heads/golden/crossnode-294  (new branch, nothing overwritten)
VERIFIED BEFORE ANY RUN ID WAS READ, and recorded at that instant:
  ls-remote sha = a9e786b24fe4e541dcd394ef428f225d72f3024a  OK
  remote tree   = 74fa640f78b0a588cd01b90c5898f9ea73e5dc08  OK
Both match the values I had recorded pre-push, which were themselves derived independently of doyle's.

RUN 34445961595, event push, head_sha correct, run_attempt 1, run_started_at 06:36:27Z.
PR: BigscreenVR/spt-bs-core#218, assembly/v0.69.0 -> main.

SCORING AGAINST NINE ROWS, and the correction has already paid for itself. Six rows exist so far and TWO OF THEM ARE n1-gate — the job my previous commune's "six job keys" list omitted entirely. Had I scored this run against the number I was carrying, both n1-gate rows would have read as rows I could not account for. Current state:
  traceability                              success
  changes                                   success
  n1-gate  (Linux, kitsubito)               success
  n1-gate  (Windows, hfenduleam)            in_progress
  test     (Linux, kitsubito)               queued
  test     (Windows, hfenduleam)            queued
  twohost-b / twohost-a / notify            not yet created (needs-gated)
Full expected set: changes, test x2, n1-gate x2, twohost-b, twohost-a, traceability, notify = NINE.

PR BODY CARRIES THE CORRECTED #287 RATIONALE, the rig version: heal reached only from shellwake.rs:907 as the first leg of reconcile_once (612 def, 1117/1151 past the #[cfg(test)] boundary at 1042), neither rig starts a wake host, start_inproc_daemon brings up seed control + broker-owning-hermetic-NetHost + tunnel control, plus drive control in gateway. All THREE dead rationales are named as corrected rather than silently swapped, and the body states plainly that a9e786b2's merge body still carries the false one and why it was not re-cut. doyle's two releases#294 comments are cited there.

Also in the PR body, stated rather than glossed: the head-level gate's kind(lib)+kind(bin) selection EXCLUDES the e2e binaries, so the e2e proof at this tree is hertz's and is attributed to him, and head-level e2e coverage comes from this golden lane itself.

A RED HANDS STRAIGHT BACK TO DOYLE for RCA-first triage. I do not triage it. Next word from me is the nine-row verdict.