swap v0-chat-2api and digital-human for mlx-spatial on shelf 03

Author: benjipeng

index.html

@@ -381,19 +381,14 @@ <h3 class="aa-shelf-title">Harnesses &amp; runtimes</h3>
             <p class="aa-repo-desc">HTML to Markdown for agent pipelines: a fast Go engine wrapped in a Python CLI and API, published on PyPI.</p>
             <span class="aa-repo-tag">Go</span>
           </a>
-          <a class="aa-repo" href="https://github.com/appautomaton/v0-chat-2api">
-            <span class="aa-repo-name">v0-chat-2api</span>
-            <p class="aa-repo-desc">A local OpenAI-compatible proxy for Vercel v0 chats. It serves /v1/chat/completions, keeps conversation state in SQLite, and includes a small dashboard.</p>
-            <span class="aa-repo-tag">TS</span>
-          </a>
         </div>
       </div>
 
       <div class="aa-shelf">
         <div class="aa-shelf-side">
           <div class="aa-shelf-num">SHELF 03</div>
           <h3 class="aa-shelf-title">On-device, in pure MLX</h3>
-          <p class="aa-shelf-desc">Models that run on the Apple GPU with no cloud and no PyTorch. Speech, vision, and video, all MLX-native, all on the laptop.</p>
+          <p class="aa-shelf-desc">Models that run on the Apple GPU with no cloud and no PyTorch. Speech, vision, video, and 3D, all MLX-native, all on the laptop.</p>
           <div class="aa-shelf-hint">mlx-speech tts --model fish-s2-pro --text "..."</div>
         </div>
         <div class="aa-shelf-repos">
@@ -417,9 +412,9 @@ <h3 class="aa-shelf-title">On-device, in pure MLX</h3>
             <p class="aa-repo-desc">MLX-native computer vision: object detection, segmentation, and open-vocabulary grounding with SAM 3 and LocateAnything.</p>
             <span class="aa-repo-tag">Python</span>
           </a>
-          <a class="aa-repo" href="https://github.com/appautomaton/digital-human">
-            <span class="aa-repo-name">digital-human</span>
-            <p class="aa-repo-desc">A talking-avatar workspace where InfiniteTalk, MuseTalk, and Qwen3-TTS come together for audio-driven portrait video.</p>
+          <a class="aa-repo" href="https://github.com/appautomaton/mlx-spatial">
+            <span class="aa-repo-name">mlx-spatial</span>
+            <p class="aa-repo-desc">3D and spatial inference on device. SAM 3D Objects, TRELLIS.2, WorldMirror, and MapAnything turn images into meshes, Gaussian splats, and point clouds.</p>
             <span class="aa-repo-tag">Python</span>
           </a>
         </div>
@@ -493,12 +488,11 @@ <h2 class="aa-section-title">Three commands.</h2>
         <a href="https://github.com/appautomaton/ltx-video-mlx">ltx-video-mlx</a>
         <a href="https://github.com/appautomaton/tnt-asr">tnt-asr</a>
         <a href="https://github.com/appautomaton/mlx-cv">mlx-cv</a>
+        <a href="https://github.com/appautomaton/mlx-spatial">mlx-spatial</a>
       </div>
       <div class="aa-footer-col">
         <div class="aa-footer-head">Tools &amp; bridges</div>
         <a href="https://github.com/appautomaton/markmaton">markmaton</a>
-        <a href="https://github.com/appautomaton/v0-chat-2api">v0-chat-2api</a>
-        <a href="https://github.com/appautomaton/digital-human">digital-human</a>
         <a href="https://github.com/appautomaton">org page</a>
       </div>
     </div>