docs(plan): the plan's list of the rules added on 24 Sep no longer says none of them is built, since Track U U-32 built P4-14 above T_mem in memory mode (found with review F3.1)

Next to the P4-6a row that F3.1 corrected, the plan's "Added 24 Sep 2026"
list of P4-11 to P4-16 still ended "None is built.". Track U U-32 built
P4-14 above T_mem in memory mode on 2 Oct 2026: a native walk that fails
part-way above T_mem ends the scan (noRestartRefusal, which startScan
throws). docsTruth's P4-14 test already holds the roadmap and DESIGN to that.
The old sentence is now struck. In its place: none was built on
24 Sep 2026, since then U-32 has built P4-14 above T_mem in memory mode, and
each task's entry records what it built. The P4-14 test now reads this
paragraph as well.

Mutants, all red: the old sentence restored, the new clause dropped, the old
sentence left unstruck, and "in memory mode" dropped. Every file was
restored, checked by SHA-256.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Author: Prithvi-Web

docs/superpowers/plans/2026-09-18-phase4-storage.md

@@ -41,7 +41,7 @@
 | P4-9 | **Incremental rescan for the native engine is not built in this phase, with the measurement recorded:** with one listing call per directory, revalidating every directory's own `(dev, ino, mtime)` costs the same syscalls as listing it, so an mtime-keyed index cannot be faster than the walk (the prompt's premise was one `stat` per file). The legacy walker keeps its mtime cache. A change-journal rescan (FSEvents / USN) is the only faster path and is recorded in DESIGN §9.1 as deferred, with this reason, for Phase 8's decision. `cacheHitRate` in the stats stays `null` with the reason in `engineReason`. | Honest numbers; no feature that cannot be measured to help. |
 | P4-10 | **Every memory number comes from `bench/`** (peak RSS of the scanning child process, the mode, the entry count, this machine's tier), recorded under `bench/baselines/` with `--record` only on a clean tree. | Phase 1's rules. |
 
-**Added 24 Sep 2026 — see §S.0:** P4-11 (every large Rust allocation is an anonymous mapping, freed with `munmap` (`VirtualFree` on Windows)), P4-12 (the main thread sees only `PackedScanStore`; `NativeScanStore` only in worker threads), P4-13 (the job queue is FIFO up to `Q_MAX`, LIFO beyond), P4-14 (no legacy fallback in spill or aggregate or above `T_mem`), P4-15 (one availability table) and P4-16 (exact, or flagged, or refused). None is built.
+**Added 24 Sep 2026 — see §S.0:** P4-11 (every large Rust allocation is an anonymous mapping, freed with `munmap` (`VirtualFree` on Windows)), P4-12 (the main thread sees only `PackedScanStore`; `NativeScanStore` only in worker threads), P4-13 (the job queue is FIFO up to `Q_MAX`, LIFO beyond), P4-14 (no legacy fallback in spill or aggregate or above `T_mem`), P4-15 (one availability table) and P4-16 (exact, or flagged, or refused). ~~None is built.~~ None was built on 24 Sep 2026; since then Track U U-32 has built P4-14 above `T_mem` in memory mode (2 Oct 2026), and each task's entry below records what it built.
 
 ---
 

tests/docsTruth.test.ts

@@ -241,7 +241,7 @@ test('ULTRA-ROADMAP says how the Track U gate stands for every release granted a
 const paragraphs = (doc: string): string[] =>
   doc.replace(/~~[\s\S]*?~~/g, '').split(/\n\s*\n/).map((p) => p.replace(/\s+/g, ' ').trim());
 
-test('ULTRA-ROADMAP and DESIGN say P4-14 is built above T_mem in memory mode (Track U U-32), as startScan does (found with review F1.5; DESIGN, review F2.2)', () => {
+test('ULTRA-ROADMAP, DESIGN and the Phase 4 plan say P4-14 is built above T_mem in memory mode (Track U U-32), as startScan does (found with review F1.5; DESIGN, review F2.2; the plan, found with review F3.1)', () => {
   // The code: a native walk that fails part-way above T_mem is refused a
   // walk again in the built-in walker (noRestartRefusal; startScan throws it,
   // the README test above). The roadmap still called P4-14 "decided, not
@@ -270,6 +270,12 @@ test('ULTRA-ROADMAP and DESIGN say P4-14 is built above T_mem in memory mode (Tr
   assert.match(rule, /\(P4-14, decided by the owner on 28 Sep 2026, plan §S\.11 Q10\)/, 'the rule is the owner\'s decision of 28 Sep');
   assert.match(rule, /Track U U-32 built it above `T_mem` in memory mode/, 'built above T_mem in memory mode');
   assert.match(rule, /spill and aggregate are not built/, 'and the modes it also covers do not exist yet');
+
+  // The plan's list of the rules §S.0 added on 24 Sep 2026 said "None is built".
+  const added = paragraphs(read('docs', 'superpowers', 'plans', '2026-09-18-phase4-storage.md')).filter((p) => p.startsWith('**Added 24 Sep 2026 — see §S.0:**'));
+  assert.equal(added.length, 1, "the plan lists the rules §S.0 added on 24 Sep 2026");
+  assert.doesNotMatch(added[0], /None is built/, 'not "None is built"');
+  assert.match(added[0], /None was built on 24 Sep 2026; since then Track U U-32 has built P4-14 above `T_mem` in memory mode/, 'the plan says U-32 built P4-14 above T_mem in memory mode');
 });
 
 /* ─────────────── the Phase 4 plan's questions: nothing the owner decided is called pending ─────────────── */