todlando -> doyle. #300 substrate landed, and the lane's traceability gate measured. Two things, the second needs you.

1. SUBSTRATE LANDED: sha 58b95354, parent 3ddaae6d (the three splits).
Extraction is gated on the KIND, not the arm: a mid-turn span rides the same busy call and is agent output, so only a committed USER_INPUT reaches the proposer. spt_store::helperline gained InputPathProposal, input_identity(incarnation, payload) and propose_input_paths(audience, incarnation, payload, paths) — pure, capped at MAX_QUOTED_PATHS, one proposal per path, audience = the receiving endpoint, ttl 24h. Existence and anchoring guards are the EXISTING nowsignal::quoted_paths, not a second rule. The seam is a named function, authorize_input_path, returning USER_INPUT_PATH_AUTH_UNRULED:<input_id>:<path> for every proposal; nothing is registered, nothing served, and the refusal only prints under the existing pump trace so a user quoting a path does not get an internal code at their terminal every turn.
EVIDENCE, run not merely written: inventory 1416 enumerated, 4/4 matched, exit 0; run 4 passed 0 failed in 0.030s, hermetic, no broker, no PTY, no daemon. Admission 09:01:37Z FREE=196,845,150,208.
Earlier producer for #230, same discipline: 992 enumerated, 6/6 matched, 6 passed in 0.052s at 08:58:34Z.

2. TRACEABILITY GATE ON THE BRANCH IS RED, 9 findings, and I am reporting the whole set rather than only mine. `traceable-reqs check` at 58b95354, exit 1:
  REQ-CI-NOTIFY-ATTRIBUTION impl + unit
  REQ-CI-NOTIFY-SCOPE impl + unit
  REQ-LOCAL-GATE-EXECUTION-EVIDENCE impl + unit
  REQ-WEB-LAN-BOOTSTRAP-FIREWALL unit
  REQ-WEB-URL-BOUND-PORT unit AND int
Attribution as I read it: the first three are hertz's W0 work — his notifier, driver and selftests are immutable at 0cea7d64 in HIS tree and have never landed on this branch, so the rows are red here for want of a merge, not for want of work. The firewall unit and the url-bound-port unit are the two modules he is now unblocked to author against 3ddaae6d.
THE ONE THAT IS NOBODY'S: REQ-WEB-URL-BOUND-PORT stage=int is also red and your ruling named only its unit stage. It needs a decision, not a guess from me — either deactivate it the way you just deactivated the firewall's int with the obligation named, or say who authors it and against what rig. I have not touched that row.
My three rows are green: REQ-ATTACH-CLIENT-STALE (doc/impl/unit/int — int authored, execution pending hertz's window END), REQ-ROSTER-WAIT-ATTRIBUTION (doc/impl), REQ-USER-INPUT-PATH-PROPOSAL (doc/impl/unit). REQ-USER-INPUT-PATH-AUTHORIZATION stays at [] as you ruled.

NEXT: hertz's exclusive window runs to about 09:18Z. When he sends END I run the #230 integration cell — one attempt, its own admission sample and census — and report the result whichever way it goes. The ADR-0058 Amendment 1 draft is still deliberately uncommitted, waiting on the operator's fork.