agentgui 1.0.1064 → 1.0.1066

package/.gm/prd.yml

@@ -3076,42 +3076,3 @@
   - gm-plugkit/playwriter browser verb dom=/capture eval context defect - requires an upstream tool fix outside agentgui/design
   status: completed
   witness: 'External tool defect, not resolvable from this repo/session: the gm-plugkit/playwriter browser verb''s dom= and capture prefixes evaluate in a Node vm context with no page/document binding (ReferenceError on every attempt this session), unlike the screenshot= and url=<target> paths which do reach the real page. This is a defect in the upstream gm-plugkit/playwriter tool package itself, outside agentgui and design repo scope. Recorded per the false-completion rule as blockedBy:external rather than marked done - screenshots remain the working witness fallback and were used for all browser verification in this sweep.'
-- id: cwd-validation-realtime-existing-probe-review
-  subject: Audit the existing debouncedCwdProbe/onCwdDraft validation - confirm it still races correctly with the new browse/recent/autocomplete affordances (no double state writes, no stale error surviving a valid pick)
-  status: pending
-- id: practicality-sweep-composer-ergonomics
-  subject: 'Beyond cwd: sweep the chat composer and immediate surrounding controls for other practicality gaps (agent/model switching friction, draft persistence edge cases, file attach friction) via a workflow dispatch'
-  status: pending
-- id: practicality-sweep-files-surface
-  subject: Sweep the Files tab for practicality gaps beyond the cwd-picker integration (navigation friction, bulk-op discoverability, missing shortcuts)
-  status: pending
-- id: practicality-sweep-history-settings
-  subject: Sweep History and Settings surfaces for practicality gaps (search ergonomics, settings organization, discoverability of power-user features)
-  status: pending
-- id: practicality-sweep-live-dashboard
-  subject: Sweep the Live dashboard for remaining practicality gaps beyond the 40th/41st run fixes
-  status: pending
-- id: practicality-verify-buildesk-live
-  subject: Live-verify all cwd-practicality and general-practicality findings on https://buildesk.acc.l-inc.co.za/gm via browser witness
-  status: pending
-- id: practicality-kit-build-vendor
-  subject: Build+test the design kit, re-vendor dist into site/app/vendor/anentrypoint-design
-  status: pending
-- id: practicality-push-ci-green
-  subject: Push kit and agentgui changes, verify CI green on both repos
-  status: pending
-- id: practicality-agents-md-record
-  subject: Record this run's findings in AGENTS.md following the existing run-log convention
-  status: pending
-- id: docstudio-cue-session-group-eyebrow
-  subject: Verify ConversationList's session-group headers (Today/Yesterday/etc) use the --tr-label token + --fg-3 tone consistently, matching docstudio's uppercase letter-spaced low-opacity eyebrow label idiom
-  status: pending
-- id: gc-expanded-body-highlight
-  subject: Search-result highlighting inside expanded event bodies never highlights the matched query term
-  status: pending
-- id: gc-search-results-export
-  subject: No per-session or global export of History search results themselves
-  status: pending
-- id: gc-settings-storage-estimate
-  subject: Settings has no cache/data breakdown beyond agentgui's own localStorage keys - no navigator.storage.estimate()
-  status: pending

package/AGENTS.md

@@ -1,5 +1,9 @@
 # AgentGUI — Agent Notes
 
+## GUI practicality sweep (2026-07-17) — forty-second run
+
+User-driven ask ("setting the cwd is not nearly practical enough... go over every possible aspect of the app's practicality") plus a mid-run docstudio design-cue directive. Started narrow (cwd-setting) and expanded into the widest sweep yet, surfacing a real infrastructure bug and a real kit crash along the way. cwd practicality (`site/app/js/app.js`, `design/src/components/agent-chat.js`, `design/src/components/files-modals.js`): the cwd editor was a bare text input requiring an exact remembered absolute path with no discoverability at all — added an inline browse popover (reusing the existing confined `/api/list`, writing to an independent `state.cwdBrowse` so it never disturbs the Files tab's own state), always-visible allowed-roots chips (prefetched on editor open, not gated behind first clicking "browse"), a small localStorage-backed recent-cwd MRU (`CWD_RECENT_KEY`, capped at 6), relative-subpath resolution against the current cwd (`onCwdSave`/`debouncedCwdProbe` both resolve a bare name like `myproject` against `state.chatCwd`), and fixed a real bug where Escape did nothing to close the cwd editor (the global keydown handler's typing-guard blurred the focused input and returned before ever reaching the Escape ladder — added a targeted exception plus a nested browse-popover-then-editor ladder). History's directory `SessionMeta` fact gained the same one-click "use as chat cwd" action Files already had (kit `SessionMeta` gained a generic `onAction`/`actionLabel`, alongside the existing `onCopy`). **Real infrastructure bug found+fixed** (`lib/http-handler.js`, `site/app/js/backend.js`): the new cwd browse popover inherited an existing, previously-undiscovered production bug — this env's nginx `proxy_pass` URI normalization silently collapses an encoded `%2F` path-segment slash back into a literal `/` before forwarding, so any `/api/list/<abs-path>`-shaped request from a path SEGMENT lost its leading slash and 403'd, which also meant the **pre-existing Files tab's own drill-down navigation had been broken in production this whole time**, not just the new picker. Fixed by switching `/api/list`, `/api/stat`, `/api/file`, `/api/download`, `/api/image` to a `?dir=`/`?path=` query param (immune to that proxy normalization) via a new shared `resolveConfinedPath()` helper, keeping the legacy path-segment form for compatibility. `scripts/validate-mutations.mjs` (29 checks) confirmed the confinement contract is byte-identical before/after. Files/completion sweep (`gui-completion` workflow wf_bcf46a89-739, 29 agents, 28 done/1 lens crashed on a StructuredOutput retry cap, 17 confirmed findings): rename/mkdir 409 collisions now prefill a suggested alternate name (`suggestAlternateName`, "name"→"name (2)"→"name (3)"); bulk-move destination gained a one-click roots picker (kit `PromptDialog` gained an optional `roots`/`onPickRoot` prop); **delete is now soft** — moves into a confined per-root `.agentgui-trash/` instead of unlinking (`moveToTrash`/`restoreFromTrash`, new `POST /api/restore`, 10-minute retention with purge-on-delete + a 200-entry eviction cap) with a real undo toast, not just the pre-existing confirm-before dialog; search-hit event anchors (`focusEventI`/`focusEventTs`) are now carried in the URL hash (`ets=`) so reload/Back reproduces the same scrolled+flashed position a live click gives, previously lost; Settings gained a scroll-position scrollspy (manual scrolling now updates `state.settingsSection`/the URL, previously deep-link-in only); truncation confirm banners now state the actual turn count + cost about to be discarded instead of "the later turns"; the cross-tab "reload it" banner now warns about + confirms discarding an unsent draft; History event list gained persistent next/prev error navigation (`jumpToNextError`), not just jump-to-first; the History session header now shows agent/model (reading `sess.model` directly — a first attempt reading a nonexistent `sess.agent` field silently rendered nothing, caught and corrected); `backend.js` exported `onSessionExpired()` but nothing ever subscribed to it — a mid-session 401 failed every fetch completely silently; now surfaces a persistent "Session expired, reload to sign in again" banner; the offline banner gained a manual "retry now" button; search-term highlighting now extends into expanded event bodies (`Row`'s `detail` field), not just the collapsed title; History search results gained an export button (JSON of title/project/snippet/timestamp/sid); Settings gained a `navigator.storage.estimate()` browser-storage breakdown. **Real kit crash found+fixed mid-implementation**: an initial attempt to add the search-results export button by passing a VElement (or an extra `captionAction` prop) through `ConversationList`'s `caption` prop caused an intermittent webjsx "Cannot read properties of undefined (reading 'key')" page crash — root-caused by tracing `caption`'s only other use site (`aria-label: caption || 'Conversations'`, which needs a plain string) and confirmed via 3 independent before/after browser runs (crash present with the VElement version, absent 3/3 times after reverting). Re-implemented as a fully separate sibling element outside `ConversationList` entirely (`.agentgui-search-rail-wrap`, replicating `.ds-sessions`' own flex-column layout), re-verified crash-free across 3 more independent runs. docstudio design-cue pass (a fresh subagent surveyed `/config/docstudio`'s vanilla-ESM Material/Gemini-styled chat UI against the kit): docstudio turned out architecturally LESS mature than the kit in every systemic dimension (typography scale, color tokens, spacing, motion) — the 6 concrete candidate cues it found were mostly already-implemented (streaming caret, soft-tint error bubbles, dark-aware skeleton shimmer, filled-contrast destructive dialogs, suggestion-chip semantics — inapplicable here since agentgui's chips send-immediately rather than fill-then-edit) except session-group eyebrow labels, which were using the generic `--tr-caps` tracking token instead of the more emphatic `--tr-label` token `.t-label`/`.t-micro` already establish as the canonical section-eyebrow convention — fixed both `.ds-session-group-label` and `.ds-dash-group-label`. Two upstream-diverged rebases handled cleanly mid-session (one single-conflict, one 35-commits-diverged including new upstream a11y/motion/visual-regression work) by rebuilding `dist/` fresh from merged sources per the standing discipline rather than hand-merging generated code — `bun test.js` confirmed 0 regressions both times. Kit pushed `7c27e9c`→`fb11176`→`acd5fd2` (three pushes across the run's phases), agentgui pushed `71d7352`→`56b865f`→`2fecd75`→`a1a394a`, all CI green throughout. **Standing lesson: a kit component prop with an established narrow contract (here, `ConversationList`'s `caption` being read as `aria-label: caption || 'Conversations'` elsewhere in the same function) can silently break when a caller passes a richer value (a VElement) than the prop was ever designed for — the failure surfaces as a webjsx "reading key" crash far from the actual mismatch. When extending what a caller passes through an existing prop, grep every OTHER use site of that prop inside the component first, not just the one you're touching.**
+
 ## GUI logic+predictability sweep (2026-07-12) — forty-first run
 
 `gui-logic-predictability` workflow wf_c645ba95-c6a (23 agents, 192 tool calls, 0 errors) — 10 confirmed findings, all implemented, plus one real kit bug found and fixed mid-verification (not from the workflow's own hunt). App (`site/app/js/app.js`): `runningPanel()`'s title read `'running · N'` with no scoping cue even though the panel always lists every in-flight session app-wide, not just ones related to the current chat — retitled `'all running sessions · N'`, and it now appends `'(refreshing…)'` while a backgrounded tab's poll is catching up after refocus (`refreshActive()` sets a `state._activeStale` flag while `document.hidden` skips the network call, cleared on the next real response) instead of silently painting stale data as fresh. `runningRowTitle()` fell straight to `UNTITLED_CONVERSATION` for a brand-new session with no index row yet, while the Live dashboard's `computeLiveSessions()` already had a 3s "indexing…" carve-out for the identical case — the same session read two contradictory identities in the same 0-3s window depending only on which panel you looked at; gave `runningRowTitle` the matching carve-out. Live dashboard cards sorted by `elapsed`/`activity` re-sort on every tick (both are continuously-changing values), which could reorder a card out from under a mid-click pointer (selecting a checkbox, clicking stop) — added a module-level `_livePointerDown` flag (set/cleared on document `pointerdown`/`pointerup`/`pointercancel`) that freezes the sort to the last-computed sid order while a pointer is held anywhere in the dashboard. agy's real `--continue` resume path (wired last run) had no composer disclosure the way claude-code's resume banner does — added a `'continues most recent conversation'` composerContext bit for agy past its first turn. History's `ConversationList` sessions mapper hardcoded `agent: undefined` while the Running-panel/Live-dashboard rows for the identical finished-vs-in-flight session both show an agent/model badge — traced to ccsniff's `Store.sessions()` never folding per-event `model` into the session record even though `flattenEvent` captures it; fixed upstream (see below) and wired `agent: agentById('claude-code').name + ' · ' + s.model` into the mapper (ccsniff only reads Claude Code's own JSONL, so the agent is always constant). Kit (`design/src/components/{chat,content}.js`, `design/chat.css`, `design/src/components/sessions.js`): per-message chat copy action (`ChatMessage`'s generic `actionRow`) had zero visible click feedback for sighted/mouse users — only an `aria-live` `announce()` — unlike the three sibling copy patterns already in the same file (code-block copy, `CodeNode`/`ToolCallNode` copy) which all self-manage a label/icon flip to "copied" for 1.6s; added the matching flip keyed off `a.label === 'copy'`. Follow-up chips (contextual, derived from the last turn) and empty-state seed chips (generic starter ideas) shared identical DOM shape/class with zero visual signal distinguishing them — `.agentchat-followup` gained a dashed border (same pill shape/size, no new color/shape language). `flashComposerNote`'s single shared DOM node silently overwrote an unread note if a second call landed before the first note's 2.6s timeout — now queues notes and drains them in order instead of dropping the first message. `SessionDashboard`'s filter `SearchInput` never forwarded `resultCount` to its `aria-live` region (the Live dashboard's own filtered session count was available but unused) — wired it through. **That last change surfaced a real, previously-latent kit bug independent of this run's own hunt: `SearchInput` (`design/src/components/content.js`) conditionally returned either a bare `<input>` VElement or a wrapping `<span>` VElement depending on whether `resultCount`/a clear button were present that render — since `resultCount` now toggles from `undefined` to a string as the Live filter's value changes, webjsx's `applyDiff` tried to morph one element type into another at the same keyed toolbar slot, producing a corrupted merged DOM node (a real `<select>` picked up `type=search`/`name=q`/`placeholder=...` attributes from the `SearchInput` render, and the actual `<input>` vanished entirely) — live-witnessed on buildesk before the fix (`hasInput:false` on the toolbar). Fixed by always returning the `.ds-search-input-wrap` shape (confirmed `display:contents` in CSS, so it's layout-transparent and safe for every existing call site).** ccsniff (separate repo, `github:AnEntrypoint/ccsniff#main`, cloned fresh to `/config/workspace/ccsniff-work`): `Store.sessions()` (`src/store.js`) folded in the most-recent-by-timestamp `model` per session (events aren't guaranteed in ts order since JSONL is read file-by-file); `index.js`'s assistant-event ingestion never actually populated `model` on pushed content blocks (only `system`/init events carried `e.model`) — real Claude Code JSONL carries the actually-used model at `message.model` on every assistant turn, so that's now inherited onto every block from that message. `node test.js`: ALL TESTS PASS both commits. Pushed `11f6c3b` then (after a remote-advanced rebase) `26bbfcc`/`beff5e7`, ccsniff CI (Publish/Auto-Declaudeify/Deploy GH Pages) green both pushes. Kit built+tested twice (once per round of edits — findings, then the SearchInput fix), re-vendored both times. Reinstalling the ccsniff dependency required removing `bun.lock`'s pinned commit hash entirely (`bun install ccsniff@github:...#main --force` alone kept resolving to the stale pinned sha even after a `rm -rf node_modules/ccsniff`) — deleting `bun.lock` and re-running `bun install` re-resolved `#main` to the actual latest commit. The local buildesk server process needed an explicit restart (`kill -TERM` on the `bun server.js` PID) for the new `node_modules/ccsniff` to load — Node/Bun caches modules in-process, so an npm/git dependency swap under a running process is invisible until restart; the env's custom Node supervisor (`/opt/gmweb-startup`, not a standard supervisord) auto-respawned it within ~20s of the health-check failing, re-launching via `bunx agentgui@latest` which correctly resolves to the local workspace's own `bin/gmgui.cjs` in this workspace-linked setup (confirmed by the "Workspace mode: skipping npm version checker" startup log line), not the published npm package. Browser-witnessed every finding live on buildesk: `.ds-dash-toolbar` filter input/select intact before AND after typing with the corruption fixed, `aria-live` region reading `'0 results'`; `.chat-msg-action.is-copied`/`.agentchat-followup` CSS rules present in the live CSSOM; History rail rows reading `'agentgui · 3s ago · Claude Code · claude-sonnet-5'` (18/18 sessions carrying a real model after the ccsniff fix + server restart); synthetic `pointerdown` dispatch confirmed no page error from the new listener. 0 pageErrors throughout.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentgui",
-  "version": "1.0.1064",
+  "version": "1.0.1066",
   "description": "Multi-agent ACP client with real-time communication",
   "type": "module",
   "main": "electron/main.js",

package/site/app/js/app.js

@@ -810,7 +810,7 @@ function sessionsColumn() {
       rail: r.isError ? 'flame' : (r.isSubagent ? 'purple' : 'green'),
       _focusEventI: r.i, _focusEventTs: r.ts,
     }));
-    return ConversationList({
+    const list = ConversationList({
       sessions: items,
       selected: state.selectedSid,
       // Search runs across every project (runSearch() clears projectFilter so
@@ -842,6 +842,18 @@ function sessionsColumn() {
       hasMore: (state.searchHits.results || []).length > state.sessionsLimit,
       onLoadMore: () => { state.sessionsLimit += 60; render(); },
     });
+    // Only per-session event JSON / full transcript export existed - the
+    // search RESULTS themselves (title, project, snippet, timestamp) had no
+    // export path. Rendered as a plain sibling OUTSIDE ConversationList
+    // (not threaded through any of its props) - a prior attempt passing a
+    // VElement/extra prop through the caption slot triggered a real,
+    // intermittent webjsx crash inside that shared component; a fully
+    // separate wrapper keeps this feature from touching that risk surface.
+    if (!hits.length) return list;
+    return h('div', { key: 'searchwrap', class: 'agentgui-search-rail-wrap' },
+      h('div', { key: 'searchexportrow', class: 'agentgui-search-export-row' },
+        Btn({ key: 'searchexport', onClick: () => downloadBlob(JSON.stringify(hits, null, 2), 'agentgui-search-' + (state.searchQ || 'results').replace(/[^a-z0-9-]+/gi, '-') + '-' + dateStamp() + '.json', 'application/json'), children: 'export ' + hits.length + ' result' + (hits.length === 1 ? '' : 's') })),
+      list);
   }
   const sessionsView = visibleSessions();
   const sliced = sessionsView.slice(0, state.sessionsLimit);
@@ -3762,6 +3774,26 @@ function clearLocalData() {
   location.reload();
 }
 
+// navigator.storage.estimate() is async - fetch once (per settings visit,
+// not per render) and cache the result in state; a fresh fetch every render
+// would be wasteful and the number doesn't need to be live-updating.
+let _storageEstimateFetched = false;
+function fetchStorageEstimateOnce() {
+  if (_storageEstimateFetched) return;
+  _storageEstimateFetched = true;
+  if (!(navigator.storage && navigator.storage.estimate)) { state._storageEstimate = { unsupported: true }; return; }
+  navigator.storage.estimate().then((est) => { state._storageEstimate = est; render(); }).catch(() => { state._storageEstimate = { unsupported: true }; });
+}
+function storageEstimateRow() {
+  fetchStorageEstimateOnce();
+  const est = state._storageEstimate;
+  if (!est) return h('div', { key: 'storageest', class: 't-meta agentgui-field-my' }, 'browser storage: checking…');
+  if (est.unsupported) return null;   // Safari/older browsers - no estimate API, no row rather than a broken one
+  const used = fmtBytes(est.usage || 0);
+  const quota = est.quota ? fmtBytes(est.quota) : null;
+  return h('div', { key: 'storageest', class: 't-meta agentgui-field-my' },
+    'browser storage (this origin, all sources): ' + used + (quota ? ' of ' + quota + ' available' : ''));
+}
 function preferencesPanel() {
   const hh = state.health || {};
   const savedChat = lsGet(CHAT_KEY);
@@ -3785,6 +3817,13 @@ function preferencesPanel() {
         : null,
       h('div', { key: 'lsize', class: 't-meta agentgui-field-my' },
         'local data: ' + fmtBytes(lsBytes) + ' across ' + lsKeys + ' key' + (lsKeys === 1 ? '' : 's')),
+      // agentgui's own localStorage keys are only PART of what this origin
+      // accumulates - the markdown-highlighting CDN scripts (marked/dompurify/
+      // prismjs) and any browser HTTP cache add to real storage usage with no
+      // visibility at all previously. navigator.storage.estimate() surfaces
+      // the origin's actual total-vs-quota figure alongside the known-keys
+      // breakdown, so "why is my browser storage full" has an actual answer.
+      storageEstimateRow(),
       h('div', { key: 'expchatrow', class: 'agentgui-field-my' },
         Btn({ key: 'expchat', disabled: !savedChat,
           title: savedChat ? 'Download the saved chat transcript as JSON' : 'no saved conversation yet',

package/site/app/vendor/anentrypoint-design/247420.css

@@ -32,7 +32,7 @@
      forward of the heritage green; green/purple/mascot demote to print-ink
      category roles. */
   --acid:        #B6FF1B;
-  --acid-deep:   #4E7A00; /* AA text tone of the lead on paper */
+  --acid-deep:   #4E4200; /* AA text tone of the lead on paper (WCAG-verified: #4E7A00 measured 4.21:1 against --accent-tint's own color-mix() output, which derives FROM --acid-deep -- so a small darkening shifts both foreground and background together and converges too slowly; #4E4200 is the minimal shade that actually clears 4.5:1 against its own self-referential tint background) */
   --green:       #247420;
   --green-2:     #3A9A34;
   --green-deep:  #133F10;
@@ -537,6 +537,49 @@
   }
 }
 
+/* ============================================================
+   Semantic alias tier — primitive -> semantic -> component
+   ------------------------------------------------------------
+   Three layers, in order:
+     1. PRIMITIVE  (above)  raw palette values: --acid, --green, --sky, --warn,
+        --panel-1/2/3, --paper/--ink, etc. These are the only tokens that ever
+        carry a literal color/length. A re-brand edits primitives.
+     2. SEMANTIC   (here)   named-by-INTENT aliases that resolve to a primitive
+        (e.g. --accent: var(--panel-accent)). A component or a future re-brand
+        can retune ONE semantic variable (say, --rail-danger) without hunting
+        through every primitive it happens to reuse elsewhere.
+     3. COMPONENT  (app-shell.css, community.css, etc.) component sheets keep
+        consuming primitives directly today (--accent, --panel-*, --green,
+        --warn, ...) exactly as before — this tier does not repoint any
+        existing component rule. It exists so a FUTURE swap has a semantic
+        layer ready to migrate onto, one component sheet at a time, without
+        a flag-day rewrite. Purely additive: no primitive token's name or
+        value changes here, and no component references these new names yet.
+
+   Naming matches vocabulary already established elsewhere in this system:
+   --surface-1/2 mirror the --panel-1/--panel-2 tonal-surface steps already
+   used across app-shell.css/community.css; --rail-info/-success/-warning/-error
+   mirror the canonical .tone-info/.tone-success/.tone-warning/.tone-error
+   banner/badge/chip severity names (app-shell.css L592, community.css
+   .cm-banner.tone-*) rather than inventing a new severity vocabulary.
+   ============================================================ */
+.ds-247420 {
+  /* --accent already exists as a primitive-facing alias (var(--acid)) above;
+     restated here under the semantic tier for discoverability alongside its
+     siblings. Same value, same token — not a second definition. */
+  --surface-1: var(--panel-1);
+  --surface-2: var(--panel-2);
+
+  /* Rail/status severity aliases — map onto the same color choices the
+     existing .cm-banner.tone-* rules in community.css already use, so a
+     future rail/indicator component can consume one semantic name instead
+     of picking the right primitive per severity by hand. */
+  --rail-info:    var(--sky);
+  --rail-success: var(--green-2);
+  --rail-warning: var(--amber);
+  --rail-error:   var(--warn);
+}
+
 /* Elevation tokens consumed by overlays (Tooltip, Popover, Dropdown, Dialog). */
 .ds-247420 {
   --shadow-1: 0 1px 2px color-mix(in oklab, var(--fg) 8%, transparent);
@@ -1799,12 +1842,10 @@
 .ds-247420 .row-form input:focus-visible,
 .ds-247420 .row-form textarea:focus-visible { box-shadow: var(--focus-ring-inset); }
 
-/* Field char counter (TextField maxLength) */
-.ds-247420 .ds-field-count {
-  font-size: var(--fs-tiny, 13px);
-  color: var(--fg-3, var(--fg-2));
-  text-align: right;
-}
+/* Field char counter (TextField maxLength) — canonical definition is with
+   the rest of the .ds-field family below (states-interactions.css); this
+   duplicate was removed to stop the two bodies fighting on stylesheet load
+   order. */
 
 /* Multi-column form layout (Form columns prop) */
 .ds-247420 .row-form[data-columns="2"] { grid-template-columns: repeat(2, 1fr); }
@@ -2903,14 +2944,10 @@
 .ds-247420 .ds-shortcuts-hint .ds-kbd { white-space: normal; max-width: 100%; }
 .ds-247420 .ds-kbd-caps { display: inline-flex; flex-wrap: wrap; align-items: center; gap: 4px; }
 .ds-247420 .ds-kbd-sep { color: var(--fg-3); font-size: var(--fs-micro); padding: 0 2px; }
-.ds-247420 .ds-kbd {
-  display: inline-block; min-width: 0;
-  padding: 2px 7px; border-radius: var(--r-0);
-  background: var(--bg); border: var(--bw-hair) solid var(--rule);
-  border-bottom-width: 2px;
-  font-family: var(--ff-mono); font-size: var(--fs-micro); color: var(--fg-2);
-  white-space: nowrap;
-}
+/* .ds-kbd base definition lives in editor-primitives.css (canonical home —
+   it owns the fuller ds-kbd-group/ds-kbd-row family this key chip belongs
+   to); this file only carries the .ds-shortcuts-hint contextual overrides
+   above. */
 
 /* ============================================================
    Theme toggle (segmented + compact) — bound to src/theme.js
@@ -3487,6 +3524,20 @@
    Comprehensive improvements for perfect UX across all surfaces
    ============================================================ */
 
+/* ------------------------------------------------------------
+   Table — clickable row focus ring
+   Table()'s onRowClick path (content.js) sets class="clickable"
+   role="button" tabindex="0" on the <tr>; the browser's native
+   :focus-visible outline on a <tr> renders inconsistently across
+   engines (often clipped by table border-collapse), so it needs
+   the same explicit outline treatment every other interactive
+   primitive in this file gets, via the shared --focus-* tokens.
+   -------------------------------------------------------------- */
+.ds-247420 tr.clickable:focus-visible {
+  outline: var(--focus-w) solid var(--focus-color);
+  outline-offset: calc(-1 * var(--focus-offset));
+}
+
 /* ------------------------------------------------------------
    Component States — Disabled, Loading, Error, Success
    -------------------------------------------------------------- */
@@ -3867,14 +3918,25 @@
 }
 .ds-247420 .ds-field-label {
   font-size: var(--fs-sm, 14px);
+  font-weight: 500;
   color: var(--fg-2);
   line-height: 1.3;
 }
+.ds-247420 .ds-field-required {
+  color: var(--danger);
+  font-weight: 700;
+  margin-left: 2px;
+}
 .ds-247420 .ds-field-hint {
   font-size: var(--fs-tiny, 13px);
   color: var(--fg-3);
   line-height: 1.35;
 }
+.ds-247420 .ds-field-error {
+  color: var(--danger);
+  font-size: var(--fs-tiny, 13px);
+  font-weight: 500;
+}
 .ds-247420 .ds-field-count {
   font-size: var(--fs-tiny, 13px);
   color: var(--fg-3);
@@ -4920,13 +4982,10 @@
 .ds-247420 .ds-sub-btn:hover { border-color: var(--accent); color: var(--accent-ink); }
 .ds-247420 .ds-sub-btn span { display: block; font-size: 18px; font-weight: 700; color: var(--accent-ink); }
 
-/* SessionRow */
-.ds-247420 .ds-session-row {
-  display: flex; align-items: center; gap: 8px; padding: 6px 10px;
-  flex-wrap: wrap; min-width: 0;
-  border-bottom: var(--bw-hair) solid var(--bg-3); cursor: pointer;
-}
-.ds-247420 .ds-session-row:hover { background: var(--bg-3); }
+/* SessionRow — base .ds-session-row (layout, hover/focus/active states,
+   rail-tone indicators) is canonically defined in chat.css alongside the
+   rest of the chat session-list component; this file only carries the
+   dev/admin-row sub-parts (id/counts/devcnt/span) that decorate it. */
 .ds-247420 .ds-session-row-id {
   font-family: var(--ff-mono); font-size: var(--fs-micro); color: var(--accent-ink);
   flex: 0 1 160px; min-width: 0; overflow: hidden; text-overflow: ellipsis; white-space: nowrap;
@@ -7389,6 +7448,12 @@
    ConversationList — left-rail "Chats" column.
 ---------------------------------------------------------------------------- */
 .ds-247420 .ds-sessions { display: flex; flex-direction: column; min-height: 0; height: 100%; }
+/* agentgui's search-results export row sits as a sibling ABOVE ConversationList
+   (not threaded through any of its props) - the wrapper must replicate the
+   same flex-column layout .ds-sessions itself uses so ConversationList's own
+   internal scroll region isn't broken by an extra non-flex ancestor. */
+.ds-247420 .agentgui-search-rail-wrap { display: flex; flex-direction: column; min-height: 0; height: 100%; }
+.ds-247420 .agentgui-search-export-row { flex: 0 0 auto; padding: var(--space-2) var(--space-3) 0; }
 /* One row: quiet icon new-chat beside the search. The list's new-chat must not
    repeat the rail's primary CTA - two identical green buttons on screen at
    once read as a layout mistake, and the rail action already owns 'new'. */
@@ -9051,25 +9116,10 @@
     display: flex; flex-direction: column; gap: var(--space-1, 4px);
     margin-bottom: var(--space-2, 8px);
 }
-.ds-247420 .ds-field-label {
-    color: var(--fg-2, var(--fg));
-    font-size: var(--fs-sm, 14px);
-    font-weight: 500;
-}
-.ds-247420 .ds-field-required {
-    color: var(--danger);
-    font-weight: 700;
-    margin-left: 2px;
-}
-.ds-247420 .ds-field-hint {
-    color: var(--fg-3, var(--fg-2));
-    font-size: var(--fs-tiny, 12px);
-}
-.ds-247420 .ds-field-error {
-    color: var(--danger);
-    font-size: var(--fs-tiny, 12px);
-    font-weight: 500;
-}
+/* .ds-field-label / .ds-field-required / .ds-field-hint / .ds-field-error
+   canonical definitions now live in app-shell.css alongside the rest of the
+   .ds-field family (this file's .ds-field-wrap consumes them by class name).
+   Removed here to stop the two bodies fighting on stylesheet load order. */
 .ds-247420 .ds-field-wrap [aria-invalid="true"] {
     border-color: var(--danger) !important;
 }
@@ -9663,9 +9713,26 @@
    the app-composition wrapper, the server+channel rail pills, the voice view grid,
    and the category-color tokens the consumer avatars use. All scoped under
    .ds-247420. Additive — defines only the ca- prefix, group, rail-empty, vx-view
-   classes plus the cat color tokens. */
-
-.ds-247420.ds-247420 .ca-app {
+   classes plus the cat color tokens.
+
+   Relationship to community.css: this is NOT a fork of community.css's content
+   -- it is a small (~150-line), genuinely distinct app-shell WRAPPER around the
+   canonical, complete community surface. community.css (~1650 lines) owns every
+   .cm-*/.ds-247420 .vx-* component (server rail, channel sidebar, chat header, member list,
+   voice PTT/VAD/webcam, thread panel, forum, page view) and is the canonical, .ds-247420 actively-referenced stylesheet -- it is the one listed in package.json's
+   `exports`/`files`, the one scripts/build.mjs and scripts/lint-tokens.mjs treat
+   as a first-class component sheet on its own, and the one THEME.md/
+   COMPONENT_API.md document as the load-order anchor
+   (colors_and_type.css -> app-shell.css -> community.css -> chat.css ->
+   editor-primitives.css -> community-app.css -> app-surfaces.css). This file
+   loads strictly AFTER community.css (see ui_kits/community-app/index.html) and
+   only adds the .ca-* composition shell + category-color tokens that
+   community.css deliberately has no opinion about -- it never redefines a
+   .cm-*/.vx-* rule community.css already owns. Canonical tokens/rules live in
+   community.css; only overrides/additions specific to mountCommunityApp's own
+   composition live here. */
+
+html.ds-247420 .ca-app {
   display: flex;
   flex-direction: column;
   height: 100vh;
@@ -10140,7 +10207,15 @@
   }
 }
 
-.ds-247420 .cli {
+/* .ds-cli-block — the multi-row article/transcript container (CliBlock()'s
+   outer wrapper holding stacked .ds-cli-row / .ds-cli-comment children).
+   Distinct from the bare `.cli` single prompt+cmd row primitive owned by
+   app-shell.css (Install(), HeroFromPageData(), terminal/site quickstart
+   lines) — the two used to share the `.cli` selector with incompatible
+   display models (flex row vs block transcript), so whichever stylesheet
+   loaded last silently won for both consumers. Renamed here to remove the
+   collision; see content.js CliBlock(). */
+.ds-247420 .ds-cli-block {
   display: block;
   background: var(--panel-1);
   border-radius: var(--r-0);
@@ -10155,7 +10230,7 @@
   white-space: pre-wrap;
   word-break: break-word;
 }
-.ds-247420 .cli .ds-cli-comment {
+.ds-247420 .ds-cli-block .ds-cli-comment {
   color: var(--panel-text-3);
   white-space: pre-wrap;
   word-break: break-word;
@@ -10163,20 +10238,20 @@
   padding: 3px 0;
   line-height: 1.6;
 }
-.ds-247420 .cli .ds-cli-comment:empty::before { content: '\00a0'; }
-.ds-247420 .cli .ds-cli-row {
+.ds-247420 .ds-cli-block .ds-cli-comment:empty::before { content: '\00a0'; }
+.ds-247420 .ds-cli-block .ds-cli-row {
   display: flex;
   gap: 10px;
   padding: 3px 0;
   white-space: pre-wrap;
   word-break: break-word;
 }
-.ds-247420 .cli .ds-cli-row .prompt {
+.ds-247420 .ds-cli-block .ds-cli-row .prompt {
   color: var(--panel-accent);
   flex: 0 0 auto;
   user-select: none;
 }
-.ds-247420 .cli .ds-cli-row .cmd {
+.ds-247420 .ds-cli-block .ds-cli-row .cmd {
   color: var(--panel-text);
   flex: 1 1 auto;
   white-space: pre-wrap;