hertz -> doyle. RUN CLOSED, plus the authorized read-only attribution refresh.

DRIVER FINAL: DRIVER_EXIT=6. LEDGER CLEANUP=PENDING residue=2 daemon_stop_refused=1. CLEANUP_REPORT_DEADLINE expired (180s) with the driver's own note: "THE REPORT IS DUE. That is not a statement that the host is tidy." LISTENER_AFTER_CLEANUP=ABSENT on 29470.

ATTRIBUTION REFRESH at 2026-09-13T07:24:43.399Z, read-only, both pids:
  PID 26792 PRESENT created=2026-09-13T07:17:30.7408520Z parent=8816 exe=[...304-w2-repr\target\release\spt.exe] cmd=[...spt.exe daemon run --detached]
  PID 460   PRESENT created=2026-09-13T07:17:30.9149270Z parent=26792 exe=[same] cmd=["...spt.exe" daemon brain --generation 0 --start-reason cold]
  ANCHOR COMPARISON: both creation times MATCH the anchors the run recorded (26792 = ...30.740Z, 460 = ...30.914Z) to the millisecond the anchors carry. Neither is MISSING; neither reads as a reused pid on this evidence. Nothing UNREADABLE.
  PARENT LINKAGE, as linkage only: 460's parent is 26792, and 26792 is PRESENT with the same creation time - so the brain's parent row is internally consistent with the supervisor identity. 26792's parent 8816 is MISSING at this instant. Per your rule I take none of this as proof of a parent's HISTORICAL identity: a current parent pid is a pointer read now, and 8816's absence means only that nothing holds that pid at this sample.

ISOLATED HOME RECORDS (fp-home-d2-a2), existing files, not created by me:
  daemon.pid content 26792, mtime 07:17:30.912Z - the home's own record agrees with the scanned supervisor identity.
  brain.ready content {"pid":460,"generation":0,"exe_hash":"72d30901...e43349b10"} - names pid 460 AND the pinned exe hash. mtime 07:24:44Z, i.e. it was REWRITTEN during my read; the brain is live and still working.
  daemon-spawn.lock and daemon-effects.log present, both zero bytes.
  logs/daemon.stderr.log, 391 bytes at that mtime but larger now, carries the boot sequence for both roles: broker generation 0 pid 26792, BRAIN_UP pid 460 generation 0 start-reason cold. It also records, unprompted: DOCS_SERVER_BIND_FAIL on port 5474 (in use), NET_FAMILY_GATE binding IPv4-only, and an INBOUND_REACHABILITY warning that the existing spt firewall rule admits C:\Users\decid\AppData\Local\spt-core\bin\spt.exe while this daemon runs the worktree binary. I am reporting those as present in the log; none of them is a finding of mine and none is acted on.

So the disposition is unchanged and PENDING: two identities of this run, attributed by pid + creation time against the recorded anchors, still LIVE; the daemon stop refused with native exit 3 (IR-122, expected, never worked around); the listener is measured absent; no rule was created and none was removed.
Execution hold stays. No stop repetition, no signal, no kill, nothing touched. The residue-disposition decision is yours.