diff --git a/docs/BACKLOG.md b/docs/BACKLOG.md
index 981d4f1447..c2f56b3703 100644
--- a/docs/BACKLOG.md
+++ b/docs/BACKLOG.md
@@ -676,6 +676,7 @@ are closed (status: closed in frontmatter)._
 - [ ] **[B-0713](backlog/P2/B-0713-soraya-round50-imaginary-stack-toy-model-registry-gap-lean-2026-05-23.md)** Soraya round-50 hand-off — register Lean ImaginaryStack/ToyModel in verification-registry.md (sorry-bearing artifact with HaPPY-paper fidelity claim)
 - [ ] **[B-0714](backlog/P2/B-0714-soraya-round51-tla-cfg-runnability-gap-3-specs-2026-05-23.md)** Soraya round-51 hand-off — author 3 missing TLA+ `.cfg` files (AsyncStreamEnumerator / ConsistentHashRebalance / DictionaryStripedCAS) — runnability gap distinct from B-0709 registry gap
 - [ ] **[B-0715](backlog/P2/B-0715-soraya-round52-istimeinvariant-axiom-registry-gap-dbsp-chain-rule-2026-05-23.md)** Soraya round-52 hand-off — register `IsTimeInvariant` axiom in verification-registry (Class 1/2 statement+paper-drift on a load-bearing axiom that BOTH registered DBSP theorems depend on)
+- [ ] **[B-0717](backlog/P2/B-0717-soraya-round57-lsm-spine-registry-and-bp16-cross-check-pair-2026-05-24.md)** Soraya round-57 hand-off — LSM Spine cluster registry-rows + BP-16 cross-check pair (SpineAsyncProtocol candidate-P0 TLA+/code-drift gap)
 
 ## P3 — convenience / deferred
 
diff --git a/docs/backlog/P2/B-0717-soraya-round57-lsm-spine-registry-and-bp16-cross-check-pair-2026-05-24.md b/docs/backlog/P2/B-0717-soraya-round57-lsm-spine-registry-and-bp16-cross-check-pair-2026-05-24.md
new file mode 100644
index 0000000000..4f32c2aaa3
--- /dev/null
+++ b/docs/backlog/P2/B-0717-soraya-round57-lsm-spine-registry-and-bp16-cross-check-pair-2026-05-24.md
@@ -0,0 +1,103 @@
+---
+id: B-0717
+priority: P2
+status: open
+title: "Soraya round-57 hand-off — LSM Spine cluster registry-rows + BP-16 cross-check pair (SpineAsyncProtocol candidate-P0 TLA+/code-drift gap)"
+created: 2026-05-24
+last_updated: 2026-05-24
+classification: buildable-now
+decomposition: atomic
+assignee: kenji
+discovered_by: soraya
+owners: [kenji, formal-verification-expert]
+type: cross-check-gap
+composes_with:
+  - docs/backlog/P2/B-0709-soraya-registry-coverage-drift-register-11-unregistered-specs-2026-05-23.md
+  - docs/backlog/P3/B-0716-soraya-round53-b0709-scope-correction-3-lsm-spine-specs-2026-05-23.md
+  - tools/alloy/specs/Spine.als
+  - tools/tla/specs/SpineAsyncProtocol.tla
+  - tools/tla/specs/SpineMergeInvariants.tla
+  - docs/research/verification-registry.md
+---
+
+# B-0717 — LSM Spine cluster: registry-rows + BP-16 cross-check pair (Soraya round-57 hand-off, post-B-0716-merge)
+
+## Origin
+
+Soraya's ninth autonomous routing tick (2026-05-24 — round 57, immediately post-merge of PR #4791 / B-0716 P3 scope-correction at 00:06:45Z).
+
+Re-engagement trigger fired per rounds 54+55+56 named conditions. Soraya re-engaged at trigger-firing and surfaced substantive routing follow-up that B-0716 didn't fully cover.
+
+## Finding
+
+B-0716 amended B-0709's enumeration from 11 → 14 unregistered specs (added LSM Spine cluster: `Spine.als` + `SpineAsyncProtocol.tla` + `SpineMergeInvariants.tla`). PR #4791 MERGED the backlog enumeration correction.
+
+**The registry-row execution payload is still pending.** `docs/research/verification-registry.md` has **zero `Spine` matches** as of this row's authoring. The 3 LSM Spine specs are still unregistered.
+
+Additionally, Soraya's round-57 BP-16 cross-check triage on the Spine cluster identified a **candidate-P0 TLA+/code-drift gap** on `SpineAsyncProtocol.tla` that warrants paired-tool coverage with FsCheck.
+
+## Two bounded subitems (separable; both ship within B-0709 umbrella)
+
+### Subitem (a) — Registry-row capture for the 3 Spine specs
+
+Continuation of B-0709 execution payload. Same shape as the 11 already-registered specs.
+
+Per-spec routing (confirmed correct on-disk):
+
+| Spec | Property class | Primary tool | Routing |
+|---|---|---|---|
+| `tools/alloy/specs/Spine.als` | Structural shape (LSM levels, sorted runs, no-cycles) | Alloy | Correctly routed |
+| `tools/tla/specs/SpineAsyncProtocol.tla` | State-machine safety + concurrency (async compaction/flush interleavings) | TLA+ | Correctly routed |
+| `tools/tla/specs/SpineMergeInvariants.tla` | State-machine safety invariant (merge preserves sortedness + total-key-set) | TLA+ | Correctly routed |
+
+**Acceptance**: 3 new registry rows in `docs/research/verification-registry.md` citing O'Neil 1996 paper anchor + per-spec preconditions + audit cadence.
+
+**Effort**: S. **CI-gate**: yes (verification-drift-auditor scan should detect on next cadence).
+
+### Subitem (b) — BP-16 cross-check on SpineAsyncProtocol (candidate-P0)
+
+Routing triage by criticality:
+
+- `Spine.als` → P1 (structural drift detectable on review; single-tool sufficient)
+- `SpineAsyncProtocol.tla` → **candidate-P0** (concurrent flush/compact interleavings with merge-invariant violation = silent data corruption shape)
+- `SpineMergeInvariants.tla` → P1 (merge violation surfaces as failing read-back acceptance test; single-tool sufficient unless implementation non-trivially changes)
+
+**The wrong-tool cost if SpineAsyncProtocol stays single-tool TLA+ with no FsCheck pair**: classic TLA+/code-drift — the spec models an abstraction of the F# `Spine` implementation; if the implementation evolves (e.g., compaction-scheduling change) without TLA+/spec update, the spec still passes TLC but no longer constrains real code. FsCheck property running actual `Spine` insert/compact/read under threading would catch the drift in CI.
+
+This is the same gap class round 42 named for `InfoTheoreticSharder` pre-triple (now closed by the existing pattern).
+
+**Acceptance**: FsCheck property file `tests/Tests.FSharp/Algebra/Spine.AsyncProtocol.Properties.fs` (mirrors PR #4780's residuated-lattice analog shape — just shipped 980/980) exercising the real `Spine` F# implementation under simulated async flush/compact interleavings, asserting the TLA+ merge invariants on the executing code path.
+
+**Effort**: S. **CI-gate**: yes once added.
+
+## TLA+-hammer guard
+
+N/A for subitem (a) — registry capture, not tool routing.
+
+For subitem (b): the guard fires the **INVERSE** direction (consistent with B-0712 framing). TLA+ IS the right primary for SpineAsyncProtocol (concurrent flush/compact interleavings); but the candidate-P0 criticality argues for paired-tool coverage to close the TLA+/code-drift class specifically. FsCheck-only would miss the interleavings; TLA+-only misses the code-drift. Both compose; neither suffices alone.
+
+## Coverage impact
+
+After both subitems land:
+
+- 3 new registered specs (Spine cluster); portfolio numerator +3
+- 1 new FsCheck cross-check on candidate-P0 spec; BP-16 coverage at +1 paired-tool
+- TLA+/code-drift class closed on async compaction (silent data corruption shape prevented)
+
+## Composes with
+
+- [`docs/backlog/P2/B-0709-soraya-registry-coverage-drift-register-11-unregistered-specs-2026-05-23.md`](B-0709-soraya-registry-coverage-drift-register-11-unregistered-specs-2026-05-23.md) — umbrella
+- [`docs/backlog/P3/B-0716-soraya-round53-b0709-scope-correction-3-lsm-spine-specs-2026-05-23.md`](../P3/B-0716-soraya-round53-b0709-scope-correction-3-lsm-spine-specs-2026-05-23.md) — B-0709 scope-correction (PR #4791 MERGED 00:06:45Z); B-0717 IS the follow-up execution payload + cross-check triage
+- [`tools/alloy/specs/Spine.als`](../../../tools/alloy/specs/Spine.als) + [`tools/tla/specs/SpineAsyncProtocol.tla`](../../../tools/tla/specs/SpineAsyncProtocol.tla) + [`tools/tla/specs/SpineMergeInvariants.tla`](../../../tools/tla/specs/SpineMergeInvariants.tla)
+- [`docs/research/verification-registry.md`](../../research/verification-registry.md)
+- PR #4780 (peer Otto implementing B-0711 Residuated FsCheck — same shape as subitem (b) needs)
+- B-0710 / B-0711 / B-0712 (sibling cross-check rows from rounds 43-45)
+- `memory/persona/soraya/NOTEBOOK.md` round-57 entry (pending append)
+
+## Substrate-honest framing
+
+Subitem (a) is mechanical follow-up to B-0716 (which corrected the enumeration); B-0716 did NOT execute the registry-row authoring. Subitem (b) is genuinely-new BP-16 routing triage that B-0709/B-0716 didn't surface — it required the LSM Spine cluster to be in-scope before the criticality assessment could run.
+
+This row is filed under P2 because subitem (b) is candidate-P0 in BP-16 terms (silent data corruption shape on async compaction); the registry-row work in subitem (a) is P3-shaped but bundles cleanly with (b) under the Spine-cluster umbrella.
+
+Per Aaron's 2026-05-23 21:30Z policy-flip: Otto auto-ships immediately. Kenji owns execution.