docs: stop promising UI that doesn't exist

Sbussiso · claude · Sbussiso · commit 4f37f72b1af8 · 2026-05-08T01:13:40.000-07:00
Two pages claimed UI surfaces that never shipped. Rewrite to match
reality.

  - CameraGroups.jsx: the page told users "Open Settings &gt; Camera
    Groups, click New Group, drag cameras into the group" — but no
    such Settings section, button, or drag interaction exists. Camera
    groups are a real backend primitive (CameraGroup model + REST
    routes + MCP `list_camera_groups`) that's currently surfaced
    only to AI agents. Rewrite the page to be explicit: "Today,
    groups are MCP / API-only. A management UI inside Settings is
    planned but not yet shipped." Document the actual REST endpoints
    for users who want them now.

  - Dashboard.jsx: three false claims in the Live view section.
      * Status pill enum was "(online, offline, or stale)" — the
        actual statuses on CameraCard are online | streaming |
        recording | starting | restarting | failed | error |
        offline | suspended. No "stale".
      * "Tiles are grouped by camera group" — DashboardPage renders
        a flat list. No group_id references anywhere in the file.
      * "Click a tile to expand to full screen, or select multiple
        tiles to watch side-by-side" — no multi-select / multi-view
        mode exists. Trim the bullet header from "Fullscreen +
        multi-view" to "Fullscreen" and update the diagram alt-text
        to match. (The diagram image itself still visually shows a
        multi-view branch — needs regenerating separately.)

Co-Authored-By: Claude Opus 4.7 (1M context) &lt;noreply@anthropic.com&gt;
diff --git a/frontend/src/pages/docs/CameraGroups.jsx b/frontend/src/pages/docs/CameraGroups.jsx
@@ -1,35 +1,64 @@
-import { Link } from "react-router-dom"
-
-
 function CameraGroups() {
   return (
     <section className="docs-section" id="camera-groups">
       <h2>Camera Groups<a href="#camera-groups" className="docs-anchor">#</a></h2>
       <p>
-        Camera groups are user-defined zones — "Front yard", "Workshop", "Main floor" —
-        that bundle cameras together for filtering, display, and MCP agent navigation.
+        Camera groups are an organizational primitive — "Front yard", "Workshop",
+        "Main floor" — that bundle cameras together so AI agents (and
+        eventually a settings UI) can talk about a location instead of a
+        list of camera IDs.
       </p>
 
-      <h3>Creating a group</h3>
-      <ol>
-        <li>Open <Link to="/settings">Settings</Link> &gt; Camera Groups</li>
-        <li>Click <strong>New Group</strong>, give it a name and a color</li>
-        <li>Drag cameras into the group, or assign a group from the camera's settings drawer</li>
-      </ol>
+      <div className="docs-callout docs-callout-info">
+        <p>
+          <span className="docs-callout-icon">ℹ️</span>
+          <span>
+            <strong>Today, groups are MCP / API-only.</strong> A management
+            UI inside Settings is planned but not yet shipped. Until then,
+            groups are created and assigned via the REST API directly — see
+            below — or simply ignored if you don't have an agent workflow
+            that needs them.
+          </span>
+        </p>
+      </div>
 
-      <h3>What groups do for you</h3>
+      <h3>What groups do today</h3>
+      <ul>
+        <li>
+          <strong>MCP navigation</strong> — Agents call <code>list_camera_groups</code> to
+          resolve a natural-language location ("check the workshop") to a
+          set of <code>camera_id</code>s. This is the primary use case
+          right now.
+        </li>
+        <li>
+          <strong>Sentinel scoping</strong> — Camera scope on the Sentinel
+          page is per-camera; group-aware scope selectors are on the
+          roadmap.
+        </li>
+      </ul>
+
+      <h3>Managing groups via the API</h3>
+      <p>
+        With an admin Clerk session token, the following endpoints are
+        available:
+      </p>
       <ul>
-        <li><strong>Live view layout</strong> — The dashboard tile grid is grouped by camera group, so related cameras stay adjacent.</li>
-        <li><strong>Color tagging</strong> — Each group has a color. Tiles and tile borders reflect it so you can read a 20-camera grid at a glance.</li>
-        <li><strong>Filter and search</strong> — Filter the live view or access logs by group name.</li>
-        <li><strong>MCP navigation</strong> — Agents call <code>list_camera_groups</code> to resolve a natural-language location ("check the workshop") to a set of <code>camera_id</code>s.</li>
+        <li><code>GET /api/camera-groups</code> — list groups in your org</li>
+        <li><code>POST /api/camera-groups</code> — create a group (<code>name</code>, <code>color</code>, <code>icon</code>)</li>
+        <li><code>DELETE /api/camera-groups/&#123;id&#125;</code> — delete a group; member cameras are unassigned</li>
+        <li><code>PUT /api/cameras/&#123;camera_id&#125;/group</code> — assign a camera to a group (or pass <code>group_id=null</code> to unassign)</li>
       </ul>
 
-      <h3>Tips</h3>
+      <h3>Tips for when the UI lands</h3>
       <ul>
-        <li>Name groups by <em>place</em>, not purpose. "Driveway" stays meaningful as cameras come and go; "Vehicle monitoring" doesn't.</li>
-        <li>Use a color system your team recognizes — e.g. red for perimeter, blue for interior, green for delivery zones.</li>
-        <li>A camera can only be in one group. If you need multi-group overlap, duplicate the camera tile in a saved view instead (planned feature).</li>
+        <li>
+          Name groups by <em>place</em>, not purpose. "Driveway" stays
+          meaningful as cameras come and go; "Vehicle monitoring" doesn't.
+        </li>
+        <li>
+          A camera can only be in one group. If you need multi-group
+          overlap, that's a planned feature ("saved views").
+        </li>
       </ul>
     </section>
   )
diff --git a/frontend/src/pages/docs/Dashboard.jsx b/frontend/src/pages/docs/Dashboard.jsx
@@ -9,7 +9,7 @@ function Dashboard() {
           <source srcSet="/images/dashboard-ia.webp" type="image/webp" />
           <img
             src="/images/dashboard-ia.jpg"
-            alt="Dashboard information architecture tree. Root: Dashboard (opensentry-command.fly.dev). Four children: Live view (Camera tiles, Fullscreen + multi-view, Snapshot capture, Manual record); Settings — admin-only (Node Management, Recording Policy, Organization, Subscription, Danger Zone); Admin — admin-only (Stream Access Logs, Usage Statistics, MCP Tool Activity, System Health); Incidents (Open incidents, Evidence viewer, Markdown reports, Triage actions)."
+            alt="Dashboard information architecture tree. Root: Dashboard (opensentry-command.fly.dev). Four children: Live view (Camera tiles, Fullscreen, Snapshot capture, Manual record); Settings — admin-only (Node Management, Recording Policy, Organization, Subscription, Danger Zone); Admin — admin-only (Stream Access Logs, Usage Statistics, MCP Tool Activity, System Health); Incidents (Open incidents, Evidence viewer, Markdown reports, Triage actions)."
             className="docs-diagram-image"
             width="1920"
             height="1080"
@@ -23,16 +23,16 @@ function Dashboard() {
 
       <h3>Live view</h3>
       <p>
-        The default page after sign-in. Every camera appears as a tile with a status
-        pill (online, offline, or stale) and a live HLS player you can expand. Tiles
-        are grouped by camera group and each tile shows its node name for quick
+        The default page after sign-in. Every camera appears as a tile with a live
+        status pill (online, recording, starting, offline, suspended, etc.) and an
+        HLS player you can expand. Each tile shows its node name for quick
         identification.
       </p>
       <ul>
         <li><strong>Live Streams</strong> — HLS video served same-origin through the Command Center proxy, JWT-authenticated per viewer. Starts at the live edge.</li>
         <li><strong>Snapshots</strong> — Click the camera icon to capture a single JPEG and save it on the node. Shows up in the node's snapshots list.</li>
         <li><strong>Recording</strong> — Manual start/stop per camera. Recordings are stored locally on the node in the encrypted SQLite database.</li>
-        <li><strong>Fullscreen + multi-view</strong> — Click a tile to expand to full screen, or select multiple tiles to watch side-by-side.</li>
+        <li><strong>Fullscreen</strong> — Click a tile to expand to full screen.</li>
       </ul>
 
       <h3>Settings</h3>
