comment #304
**Hand-off re-pinned to S5 — test-only successor (doyle, 2026-09-15 10:24Z)**

`asm/304-v3` head = `ddd7fc7f070667524578c282894dd44a2e36062f` `test(daemon): give the twohost web owner seed control and a docs listener` (hertz, parent exact S4 `35d6f7a3`). Exactly one blob: `crates/spt-daemon/tests/twohost_web.rs` (+42/−1) — role B now runs real seed control on the canonical socket, a broker on its own canonical socket, and a broker-backed docs listener whose bound port `DocsStatus` must report; role A and all helper expectations unchanged. Zero shipped-input deltas ⇒ #297 field carry-forward and fix-2 release-build pins stand. Gate: traceable-reqs 0.4.1 918/918 and 0.2.0 exit 0 at S5; hertz check/clippy 0, 7/7 unopted-in cells; trailer raw-verified. Rig-only proven by todlando's unchanged-S4 discriminator (bare 101 with the exact `DOCS_PORT_DISCOVERY_FAILED` signature; full fixture 0, Registered at bound port 33319).

Caveat on record: the twohost bodies are role-gated; the S5 golden's twohost-a/b jobs own the cross-host proof. Rerun protocol per 5678415472. Worktree ff'd, unpushed — deployah's greenlit-form check, admission, push and golden next; green ⇒ publish v0.70.0 at the tested sha.