{
  "summary": "The identifiable hertz-authored historical ER correction is session-boundary test repin1648b103 (Aug23), already ancestral to S2 and fix2. The original native-presentation fix and session-sweep product fix are also landed. No retained evidence identifies the recalled earlier-today hertz RCA; do not substitute these older fixes for that missing attribution. Fix2 does not change the ER test or delivery/liveness implementation.",
  "files": [
    {
      "path": ".worktrees/hertz-304-phase-b/crates/spt/tests/er_briefing_presented_e2e.rs:758-787",
      "description": "Binding, old-row absence and loud sweep checks precede the new-session delivery assertion; current failure is the latter, not the older cross-session-ID expectation."
    },
    {
      "path": ".worktrees/hertz-304-phase-b/traceable-reqs.toml:3617-3624",
      "description": "Names session-scoped briefing and native-before-first-turn presentation as distinct historical requirements."
    },
    {
      "path": ".worktrees/hertz-304-phase-b/crates/spt-daemon/src/livehost.rs:967-993",
      "description": "S2 blame establishes pre-existing session-CAS offlining and no-session reconciliation, predating fix2 by months."
    },
    {
      "path": ".spt/preserved/hertz-attachment-NO4JJOEL/rca-source.json",
      "description": "Retained hertz RCA concerns attachment registry cross-process publication, not briefing; not evidence for the recalled ER attribution."
    }
  ],
  "architecture": "Briefing composition/session-sweep and native presentation are separate contracts. S2 already contains both historical fixes; fix2's sole production change is bootstrap_firewall/windows.rs. The observed spawn-gap liveness path is source-pre-existing, but historical provenance alone cannot establish incident causality.",
  "report": "## Commit/landing facts\nRead-only git history was run against .worktrees/hertz-304-phase-b; no builds/tests/source changes.\n\n- Exact7890ead39bb7f14ed44aaae44b0951f098ffe9ac has sole parent527cd8e8d7ef192892d5630fce10e69a6d811893. Thus S2 is the immediate ancestor, not a parallel branch.\n- `git show --format=fuller --stat527cd8e8` identifies `test: harden phase-B registry and replay fixtures`, co-authored by hertz. It touches only crates/spt-daemon/tests/restart_replay_lifetime.rs and crates/spt/tests/rc_attach_truth.rs. The message describes stabilizing synthetic registry input, cleared control stamps, pre-sibling replay evidence and generation-child cleanup—not an ER fix.\n- `git log527cd8e8 -- crates/spt/tests/er_briefing_presented_e2e.rs` contains1648b103039d4b33428b3b285d51bb97b128359f,1393bbc564fa6dd914995030a07bee1fcc459042 and133e2d66fe65388ec151ab494b301445197d2afd. This proves their ancestry to S2, and transitively7890ead3.\n\n## Identifiable old seams\n1. **Cross-session retained-row identity expectation**:1648b103039d4b33428b3b285d51bb97b128359f, Aug23, `test(er): repin briefing session boundary`, co-authored by hertz. Primary commit message: clause4 re-offers a retained briefing BY ID only within the SAME session; #208 instead supersedes an old row loudly at a NEW session and delivers that session's own briefing. This was already landed before S2. It is NOT the currently failing condition: exact source checks bound controllable=true, old row absent and ENGINE_ROOM_BRIEFING_SWEPT before panicking at782 for the undelivered new row.\n2. **Spooling without early presentation**:133e2d66fe65388ec151ab494b301445197d2afd, Aug19, `fix(er): present the engine-room briefing, do not merely spool it`, co-authored by todlando. Its primary message explicitly reports the original perch was eligible (controllable=true, online, no relay); the existing idle leg delivered only at FIRST IDLE, after the first actionable turn. Remedy: detached bounded bind-wait followed by native delivery through shared claim/renderer/release-on-miss machinery. Already landed in S2. This is not evidence that a currently INELIGIBLE/offlined perch is the same old root.\n3. **Old-session briefing supersession**: `git log527cd8e8 --grep='new session sweeps'` returns6c3efe935d329595d53a85ec6301196901d783ea, `fix(er): a new session sweeps the briefings older sessions left undelivered`. Already landed in S2; current old-row-swept precondition passed.\n\n## Fix2 touched files (exhaustive)\n`git diff --name-only527cd8e8 7890ead3`:\n- CHANGELOG.md\n- crates/spt-daemon/src/bootstrap_firewall/windows.rs\n- docs-site/src/changelog.md\n- docs/297-ENFORCEMENT-CODES-DIAGNOSIS.md\n- docs/adr/0059-lan-exposure-only-for-the-bootstrap-surface.md\n- traceable-reqs.toml\n\nTherefore ER test, broker/native delivery, livehost reconciliation and store lifecycle code are unchanged across fix2. This establishes pre-existing SOURCE, not proof that firewall timing cannot expose an existing race.\n\n## Reconciler ancestry supporting Main's investigation\n`git blame -L967,993527cd8e8 -- crates/spt-daemon/src/livehost.rs` attributes mark_offline(perch, Some(info.session_id)) at972 and its same-session CAS rationale967-969 to b7debb641566a15b5b5376c92667a30764e72943 (Jul17). Resume-custody clear973-979 dates to74412c76a5ecf2e0cfdc71eb839bf6310341484c (Jul22). The offlined.push at992 dates to03b6398b20af17aca8bdac83eb6ff351dca9274a (Jun17); log formatting at991 dates to1974dc4fc (Aug29). These are not fixes introduced by527cd8e8 or7890ead3. Whether the CAS permits the bound-but-not-published spawn gap is Main's causal trace, not established by blame alone.\n\n## Explicit missing evidence\nNo historical artifact found that identifies an **earlier-today hertz ER RCA** with its message ID, exact symptom, proposed patch or commit. Searched retained Hertz evidence directories, named RCA/source/finding artifacts and all-ref recent git messages for briefing/liveness/spawn/reconcile; recent matches do not supply such a fix. The retained named rca-source.json is the separate attachment-registry race. Consequently, cannot honestly classify the recalled today's RCA as landed or unlanded, nor assert it is the bind-before-session-publication seam. What is proven is that the old hertz Aug23 repin and prior production briefing fixes are landed; no ER correction is in S2's own commit or fix2's diff."
}