memory: snapshot — U3 released at v0.5.0, pushed and deployed

First push of this repo's history: main was 26 commits ahead of origin/main, so
v0.2.0 through v0.5.0 all reached the Gitea remote in one motion. A future
session can assume a remote exists, which no earlier one could.

Records the open defect U3 found and deliberately did not fix -- a well-formed
.marks.json with a wrong-shaped answer 500s the gallery and marks pages,
measured at 42ea67f so it predates the unit -- and marks it explicitly as
awaiting an operator decision with no issue filed, rather than letting it sit in
a detail file nobody is routed to.

No recommendation is recorded for the next unit. U6 and U7 are genuinely
independent and close different defects; the last two before a 1.0 cut are a
scope-direction call.

persistent-memory.md

@@ -28,10 +28,13 @@ _As of 2026-09-22:_
   (410 → 444), deployed and verified live, 23/23 booth pages 200, and each of
   the four verbatim booths served at exactly +46 bytes, which is
   `len(EMBED_SCRIPT_TAG)` — one append, nothing else. `87e2c53` is the unit,
-  `5c20e2f` the panel fixes. Operator approved the minor and authorised the
-  push on 2026-09-22; **this is the first push of this repo's history** — it was
-  26 commits ahead of `origin/main` before it, so every earlier tag reached the
-  remote at the same time.
+  `5c20e2f` the panel fixes, `7996fbd` the release. **PUSHED AND DEPLOYED**
+  2026-09-22 on the operator's word — and it was **the first push of this
+  repo's history**: `main` was 26 commits ahead of `origin/main`
+  (`ce598b3..7996fbd`), so `v0.2.0` through `v0.5.0` all reached
+  `git@gitea.phasefinal.com:vh/booth.git` in the same motion. `main` tracks
+  `origin/main` clean now; a future session can assume a remote exists, which
+  no earlier one could.
 - **U4 released as `v0.4.0`** (operator approved the minor on 2026-09-22).
   `c3a97c1` is the unit; the release commit carries the pre-existing fixes the
   bug-hunt panel surfaced in touched files. The tag waited for the last gate to
@@ -47,7 +50,10 @@ _As of 2026-09-22:_
   next is the operator's call. ⚠ Before starting U7, read
   `persistent-memory.d/2026-09-21-u7-section-premise-half-wrong.md`: every booth
   that actually needs navigation is FLAT, so half its premise is already known
-  to be wrong.
+  to be wrong. **No recommendation is recorded for this one on purpose** — U6
+  and U7 are genuinely independent, they close different defects, and the last
+  two units before a 1.0 cut are a scope-direction call rather than a
+  dependency one.
 - **U3's tier was MINOR and the operator approved it** (2026-09-22). The
   argument that settled it, recorded because the tie-break rule says patch: a
   capability arrived AND one left — the verbatim path gained a declared public
@@ -67,6 +73,17 @@ _As of 2026-09-22:_
   (`01M34VX0SH23Y3VC92E7GM4S70`, `01M34WAFJC3RTERFYBBZJN1SVG`,
   `01M34Y2R0RAJRSN36Q8K4KAB36`; `01M340PNVRS21HPASZT38PXQPN`,
   `01M341E9XAPZEFBSPK9HPGAM0S`, `01M343SXX27Z47C3STXXRC7M42`).
+- ⚠ **ONE DEFECT IS OPEN, FOUND BUT DELIBERATELY NOT FIXED, AND IT HAS NO
+  TRACKING SURFACE YET.** A `.marks.json` that is well-formed JSON with a
+  wrong-shaped `answer` 500s **the gallery page and the marks page** —
+  reproduced, and measured at `42ea67f` so it PREDATES U3. U3 guarded its own
+  surface (`_safe_fragments`) and left those two alone rather than widening the
+  unit; the gallery is named out of scope in the U3 contract. **The operator was
+  asked where the guard belongs and has not answered** — `_hydrate` (session
+  recommendation: one predicate, one place, every surface inherits it), per
+  render site, or the template. No issue filed. Read
+  `persistent-memory.d/2026-09-22-a-wrong-shaped-answer-500s-the-gallery.md`
+  before touching marks rendering anywhere.
 - **Two dated predictions are pending and must not be forgotten.** U5's adoption
   re-measure on **2026-09-29** (two counts, see its entry — already at 3 of 24
   announced and 2 with a `why`, all from peers told nothing), and the `.forever`