W10 design-review carry-over items. M7 lifts the medication-form
day-of-week buttons, the more-horizontal trigger, and the schedule
label-input gaps to the 44 px / space-y-2 standard used by every
other settings surface. L1 pairs every animate-spin/pulse/bounce
with motion-reduce:animate-none across the component tree so users
with prefers-reduced-motion get a still surface. L3 adds the
aria-controls + panel-id pairing to the inventory + coach-evidence
<details> blocks for parity with the Health Score provenance
accordion. L4 inserts a sr-only h4 per therapy-timeline entry so
screen-reader users can jump between medication blocks.
The Coach-Feedback admin section previously rendered four distinct top-
level structures across its loading / error / empty / data branches.
The loading branch was a thin flex-row stub without the heading, the
error branch was a p-4 alert with no card chrome at all, and only the
empty + data branches mounted the Sparkles icon + 'Coach Feedback'
title inside the bg-card border-border rounded-xl border p-6 wrapper.
The result: when the user clicked into /admin/coach-feedback the
section first measured at ~52 px (loading) and then snapped down to
its full height once the query resolved, shifting everything below it.
Mirror the canonical pattern used by SystemStatusSection: render the
outer card + heading unconditionally, and switch only the body between
spinner / alert / empty / table. The header now sits at a constant
y-offset from the section root regardless of fetch state.
Adds a regression test that asserts the 'Coach Feedback' heading and
the bg-card rounded-xl outer chrome are present in every render state.
The v1.4.22 prose rewrite shipped without a way for users to signal
whether the new tone is landing. v1.4.23 H7 closes that loop with a
light-touch thumbs row + an admin first-week dashboard.
Schema: extend `RecommendationFeedback` with a polymorphic
`target_type` column (default "recommendation" so legacy rows stay
representative) plus a nullable `reason` field. New index on
(target_type, created_at) so the aggregator can scan per-surface
buckets without a full sweep. Migration `0040_recommendation_feedback_target_type`.
Route: POST /api/insights/chat/messages/:id/feedback. Body
{ rating: "helpful" | "unhelpful", reason?: string }. Verifies
ownership (404 on cross-account messages — no existence leak),
decrypts the message body for the dedup key, and writes the row with
target_type="coach" + a metric-source string encoding the user's
active prefs (`coach:tone=warm:verbosity=default`). Per-user 60/h rate
limit. 409 on a P2002 (already_rated) so a double-click is a no-op.
Aggregator: new `buildCoachFeedbackBuckets()` slices Coach rows by
(promptVersion, tone, verbosity). The summary blob the worker writes
to AppSettings.adminAiInsightsFeedbackSummary now carries both
`buckets` (pre-existing) and `coachBuckets` (H7 addition). Targets
the same singleton row so no second worker handler is needed.
Admin view: `/admin/coach-feedback` (new section slug) renders the
Coach buckets as a small DataTable (prompt version × tone × verbosity
× helpful × not-helpful × n × helpful-rate). Helpful-rate column
tinted via Dracula tokens (green ≥80%, yellow ≥50%, orange below).
First useful question this answers: "Is the v1.4.22 prose rewrite
landing well, or did the warm tone overshoot?" If the helpful-rate
for PROMPT_VERSION 4.23.x drops below 50% within the first 100
ratings, v1.4.24 walks the persona back.
UI: `<ThumbsUp>` / `<ThumbsDown>` buttons (lucide icons, no emoji)
under each persisted assistant bubble. Streaming bubbles + refusals
+ error bubbles skip the row so users can't rate before the message
lands on disk. Localised "Thanks for the signal." after submit.
OpenAPI: spec carries the new POST route + the request schema.
Coverage now includes the H4 + H7 routes — three additions over
v1.4.22 baseline.
Tests +9 unit (aggregator Coach-bucket slicing, admin section
render-paths) plus an integration round-trip for the route.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>