Fix two stale CLAUDE.md claims the Stage 5 exit audit's own sweep missed

docs/architecture/CLAUDE.md

@@ -32,15 +32,30 @@ anything beyond orientation.
 
 **One of its four sections still carries a `Planned` subsection for a
 mechanism that doesn't exist yet** (checkpointing simulation state,
-Section 3) -- anchored to the specific roadmap task that will build it
-(TASK-034) rather than left open-ended, per the maintainer's direction
-that an unbuilt piece gets a placeholder and a backlog anchor, not
-silence or a fabricated mechanism. That task's own `docs/planning/
-roadmap.md` entry carries a matching note asking for `sequences.md` to
-be updated in the same change that lands it -- check the note still
-agrees with reality whenever that task is touched, the same "a diagram
-makes claims too" discipline this directory already applies to
-`overview.md`'s system diagram (below).
+Section 3), per the maintainer's direction that an unbuilt piece gets a
+placeholder and a backlog anchor, not silence or a fabricated mechanism.
+
+**Its anchor is no longer a task, and how that happened is the useful
+part.** This paragraph used to say the subsection was "anchored to the
+specific roadmap task that will build it (TASK-034)", with that task's
+own roadmap entry carrying a matching note asking for `sequences.md` to
+be updated in the same change. **TASK-034 landed on 2026-08-29 and
+deliberately did not build checkpointing** -- Stage 5 Completion
+Criterion 4 excludes it in as many words -- so the placeholder stayed
+accurate while its anchor pointed at a closed task, and the same pass
+left `sequences.md` with no sequence for `navier_stokes_step`, which is
+what TASK-034 *did* build. A task anchor does not cover "the task landed
+but did not build the thing" (`docs/practices.md`, "A checkable trigger
+still needs somebody to check it"). The subsection now says plainly that
+no task is assigned; whoever writes one re-reads it in the same change.
+
+Check that note still agrees with reality whenever any task named in
+`sequences.md` is touched -- the same "a diagram makes claims too"
+discipline this directory already applies to `overview.md`'s system
+diagram (below), and **a discipline this file itself failed**: the Stage
+5 exit audit re-anchored `sequences.md` and did not re-read this
+companion paragraph in the same change, which is the exact
+one-file-deep sweep it had just written a rule against.
 
 **Section 2's own `Planned` subsection (driving `simulation.step()` from
 a live render loop) was replaced with the real, built sequence 2026-08-28

planning/CLAUDE.md

@@ -54,11 +54,23 @@ each one, so a list in this paragraph would be a second, unchecked copy
 of it.
 
 Whichever are still unpopulated are so on purpose, each with a stated
-trigger in `model/entities.yaml` -- `data/releases.yaml` most
-explicitly, since
-`docs/planning/releases.md` is a sustained argument that PyFlow should
-not have a release process yet, and an empty file matching a documented
-deliberate absence is correct rather than incomplete.
+trigger in `model/entities.yaml`, and an empty file matching a
+documented deliberate absence is correct rather than incomplete.
+
+**`data/releases.yaml` used to be this paragraph's own worked example of
+that, and is now the worked example of the trigger firing.** It read
+"`data/releases.yaml` most explicitly, since `docs/planning/releases.md`
+is a sustained argument that PyFlow should not have a release process
+yet". That stopped being true on 2026-08-29, when reaching the MVP fired
+one of `releases.md`'s own three recorded triggers, that document was
+rewritten with a real process, and this file was populated with PyFlow
+0.1.0. Worth keeping rather than smoothing over: the trigger mechanism
+worked, and **this sentence was still missed by the Stage 5 exit audit's
+own documentation sweep** -- it greps for "no release process" and this
+paragraph says "should not have a release process yet", which is the
+same claim in words the grep did not match (`docs/practices.md`, "A
+stage's documentation sweep is a grep, not a diff review", whose limits
+this is an instance of).
 
 These files are exempt from the repository's usual "no empty tracked
 file" rule (`docs/planning/backlog.md` A3) -- they're data, not prose.