doyle -> deployah: IO-PARSER #22 GOLDEN HEAD HAND-OFF per the runbook (docs/golden-head-intake).

HEAD: assembly/io-parser-22 @ 594ee4bc (pushed). Chain: main e326ad4b + linear wave stack
(PR #162 W1 a8f04aff → #163 W2 769fb0a9 → #164 W3 9fc213d8 → #165 W4 e1764151 → #166 W5
9c3a129a → #167 docs b7135d66, one merge 96e32db8) + PR #168 fold (hertz reap-carrier fix,
594ee4bc). All PRs carry "fixes #12".

PRE-HAND-OFF GATE (my legs at the assembled head, exit files read):
- clippy --workspace --all-targets -D warnings: 0 at 96e32db8 AND 0 incremental at 594ee4bc.
- traceable-reqs check: 0. xtask check (docs drift + llms links): OK.
- daemon_e2e reap cell 20/20 at final head (pre-#168 rate was 2/15 — hertz RCA: test dropped the
  fire-and-forward kill carrier before the row-absence postcondition; fix keeps it alive).
- Every wave gated individually this session (battery evidence in .worktrees/gate-w1-a8f04aff
  GW1_/GW2A_/GW3_/GW4_/GW5_ files + ASM22*/ASM22B* at repo root).

GREENLIT-FORM FACTS for your verification (the check is yours, these are my inputs):
- #22 members: #12 (IO parser, founding — W1/W3/W4 + funnel), #226 (commune events — W2, both
  rulings 5/6 delivered), #233 (now-signal — W5, all seven v1 categories). NOTHING dropped,
  NOTHING added vs the greenlit form. #234 (adapter event consumption) is the fast-follow and is
  NOT in this milestone — it was never a member, so its absence is conformance, not a drop.
- Operator ruling record + wave map: IO-PARSER-22-JIT.md at repo root (committed f88aa3ee).

FIRST-EXECUTION CELLS on this golden (runbook rule — named before the run):
- crates/spt/tests: io_state_payload_e2e, shortform_dispatch_e2e, seal_shortform_e2e,
  now_signal_delta_e2e (all four NEW this milestone; each ran 3-5x in wave gates, never in CI).
- crates/spt-daemon/tests/commune_io_events_int.rs (NEW, W2).
- New unit populations: spt-proto shortform/seal + ioevent, spt-runtime io_gate, spt-store
  dispatchresults + lastmsg, spt api::delivery IO arms.
- daemon_e2e reap cell runs its FIRST CI execution with the #168 treatment.

KNOWN RESIDUALS (named, none blocking):
- Watchdog-deadline cells (spawn_truth concurrent-fresh; inject_control_wedge pair) flake under
  heavy CONCURRENT box load — all delta-tested clean this session; a golden red in one of them
  wants a same-sha rerun BEFORE triage-to-me (per-sha rerun discipline), and releases#235
  documents the spawn_truth mechanism (real product race, operator triage pending).
- IR-52 stays open (docs lane did not touch the reference generator — explicit no-touch).

Red runs hand straight back to me for triage per the runbook. Board state changes after your
golden verdict: #22 ACCEPTANCE ride via alchemy on the ff of tested sha to main.