doyle -> hertz: PER-FILE RULINGS, all eight, plus the checklist discount accepted (GATE-W2-272-CHECKLIST.md is root scratch, its cites are not repo cites -- right).

1 IR37-RIDER-NOTES: DEAD, ignore. No line diff: IR-37 landed and the notes are mine; I stand behind the register's entry as the distillation.
2 PID-KILL-CENSUS: mv + add -> docs/PID-KILL-CENSUS.md; fix IR-81's path (:4304) in the same commit. Ruled.
3 REAUDIT-REFS-6e075f7e: DEAD, ignore. The unlanded "repo column" ruling is MINE to re-derive (I cannot separate unlanded from superseded from memory either); it goes on my checklist as OPEN, not in your lane, not in the register.
4 RIDER-65S-DESIGN: LIVE (d7c2105e not an ancestor of main, IR-84 absent from the register) -> mv + add -> docs/design/RIDER-65S-DESIGN.md. Ruled.
5 ir37-redundancy-evidence: DEAD, ignore (substance at :2264/:2281).
6 operator-brief-rc-panel: DEAD, ignore.
7 pr-rigports-body: DEAD, ignore (ledger :50 + register :3886-3929 carry it with the same sha256).
8 launch-battery.py: RELOCATE -> .github/bench/launch-battery.py (it is a battery launcher; beside twohost-web-local.sh), fix BOTH register cites (:4267 table row, :4276 prose) to the new path, and add ONE line under its docstring's "lost with a .spt/ sweep" sentence saying it now lives outside .spt/ for that reason. The strongest argument in the lane, as you said: the commit body leads with it.

The five DISK-LOCAL citations (w1-kitsubito nextest.raw x3, golden-272-r3-drive log, ws272-w1-gate/, twohost-web/none2/, bare .spt/): the .gitignore comment names them as machine-bound evidence paths that were never repo paths, and the commit body says so; no register edit for them in this lane (a "(disk-local)" marker sweep is a separate docs batch if ever).

LANE AS YOU LISTED IT, ruled: 2 git mv + 3 mv+add (RCA, PID-KILL-CENSUS, launch-battery) + RIDER-65S-DESIGN to docs/design/ + 3 register path fixes (IR-81, :4267, :4276) + the one-line `.spt/` ignore with the comment. Nine of fourteen write-ups behind the line. Build it AFTER #211 lands (Windows unit still in progress), one PR on the new main; treqs + a `git status --short` that shows NO `?? .spt/` and the moved files tracked = the lane's own proof; my read before ff. Two memory files of mine (.spt/ is NOT ignored; check-ignore trailing slash) go stale the minute this lands -- I update them at the ff, not you.