diff --git a/scripts/__tests__/run-vitest-watchdog.test.mjs b/scripts/__tests__/run-vitest-watchdog.test.mjs
index 3d9f4f791b..191dbf0ec2 100644
--- a/scripts/__tests__/run-vitest-watchdog.test.mjs
+++ b/scripts/__tests__/run-vitest-watchdog.test.mjs
@@ -61,6 +61,21 @@ test("deriveBudgetMs: clamps to floor and ceiling", () => {
   );
 });
 
+test("deriveBudgetMs: shard floor pins heavy slices above the false-kill window", () => {
+  // FNXC:TestInfrastructure 2026-06-20-21:52:
+  // Regression guard for the 5min -> 15min shard-floor raise. A value whose
+  // expected×multiplier lands in the *old* un-clamped window (300s..900s) must
+  // now clamp UP to the 15min floor. 150s × 3.5 = 525s, which was returned
+  // verbatim under the old 5min floor but is below the new one. Pinning the
+  // concrete floor value here means an accidental revert to 5min fails loudly
+  // instead of silently re-tightening the engine/core slices into SIGKILLs.
+  assert.equal(CLASS_BUDGET_BANDS.shard.floor, 15 * 60_000);
+  assert.equal(
+    deriveBudgetMs({ klass: "shard", expectedDurationMs: 150_000, timingsFresh: true }),
+    CLASS_BUDGET_BANDS.shard.floor,
+  );
+});
+
 test("deriveBudgetMs: unknown class falls back to the changed band", () => {
   assert.equal(deriveBudgetMs({ klass: "nonexistent" }), CLASS_BUDGET_BANDS.changed.ceiling);
 });
diff --git a/scripts/lib/run-vitest-watchdog.mjs b/scripts/lib/run-vitest-watchdog.mjs
index 362ca0623a..1294836a0d 100644
--- a/scripts/lib/run-vitest-watchdog.mjs
+++ b/scripts/lib/run-vitest-watchdog.mjs
@@ -36,16 +36,21 @@ const MINUTE = 60_000;
  */
 export const CLASS_BUDGET_BANDS = {
   // One CI shard command (may fan out across several packages via --filter).
-  // Floor is 15min, not 5min: the heaviest shard slices (@fusion/engine and
-  // @fusion/core, each split [1/2]+[2/2]) are import- and real-git-subprocess
-  // heavy, and the committed scripts/test-timings.json undercounts that
-  // overhead (most files bucket to the 100ms floor; the snapshot total is far
-  // below current wall-clock). A "fresh" but undercounting snapshot was
-  // tightening these slices to ~340-405s and SIGKILLing healthy runs on slower
-  // CI runners — the exact false-kill the band is meant to prevent (see the
-  // deriveBudgetMs note above and KTD-2). 15min sits comfortably above the
-  // observed CI need while still bounding a true hang far under the job's
-  // 60min ceiling. Mirrors the dashboard-lane heavy-lane floor.
+  /*
+  FNXC:TestInfrastructure 2026-06-20-21:51:
+  The shard watchdog floor is 15min, not 5min. The heaviest CI shard slices
+  (@fusion/engine and @fusion/core, each split [1/2]+[2/2]) are import- and
+  real-git-subprocess heavy, and the committed scripts/test-timings.json
+  undercounts that overhead (most files bucket to the 100ms floor, so the
+  snapshot total sits far below current wall-clock). A "fresh" but undercounting
+  snapshot was tightening these slices to ~340-405s and SIGKILLing healthy runs
+  on slower CI runners (engine[2/2]=145s, core[2/2]=283s locally, both pass) —
+  the exact false-kill the floor/ceiling band exists to prevent (see the
+  deriveBudgetMs note above and plan KTD-2). 15min sits above the observed CI
+  need while still bounding a true hang far under the job's 60min ceiling.
+  The dashboard-lane floor is also 15min but for separate historical reasons;
+  the two values are not coupled and may diverge.
+  */
   shard: { floor: 15 * MINUTE, ceiling: 30 * MINUTE },
   // One local changed-file package invocation.
   changed: { floor: 2 * MINUTE, ceiling: 20 * MINUTE },