De-brand: drop Claude mentions, add source link, fix HF host suffix

README.md

@@ -11,11 +11,15 @@ short_description: Calibrated in-browser facial expression reader
 
 # Live Expression Reader
 
-Real-time facial-expression analysis with **calibrated confidence**, **literature-grounded cognitive states** (tired / focused / bored / stressed / engaged / confused / calm), and **FACS-grounded AI explanations** via Claude Haiku 4.5. Runs entirely in your browser — webcam frames never leave your device.
+Real-time facial-expression analysis with **calibrated confidence**, **literature-grounded cognitive states** (tired / focused / bored / stressed / engaged / confused / calm), and **AI-powered explanations** grounded in the Facial Action Coding System. Runs entirely in your browser — webcam frames never leave your device.
 
-> **Live demo:** <https://arjun10g-live-expression-reader.hf.space>
+[![Live demo](https://img.shields.io/badge/Live%20demo-arjun10g--live--expression--reader-indigo?style=for-the-badge)](https://arjun10g-live-expression-reader.static.hf.space)
+[![Source on Hugging Face](https://img.shields.io/badge/Source-Hugging%20Face%20Spaces-orange?style=for-the-badge&logo=huggingface)](https://huggingface.co/spaces/arjun10g/live-expression-reader/tree/main)
+[![License: Apache 2.0](https://img.shields.io/badge/License-Apache%202.0-green?style=for-the-badge)](https://www.apache.org/licenses/LICENSE-2.0)
+
+> **Live demo:** <https://arjun10g-live-expression-reader.static.hf.space>
 >
-> **Source:** <https://huggingface.co/spaces/arjun10g/live-expression-reader>
+> **Source:** <https://huggingface.co/spaces/arjun10g/live-expression-reader/tree/main>
 
 ---
 
@@ -29,13 +33,13 @@ Real-time facial-expression analysis with **calibrated confidence**, **literatur
 - **Stability sparkline** — top-1 confidence over the last ~6 s.
 - **Cognitive states** computed from PERCLOS (Wierwille 1994), blink rate (Stern 1984), expression entropy (D'Mello 2010), AU4 + AU7 (D'Mello & Graesser 2010), affect-lability V-flux (Kuppens 2010 / 2013), and Russell circumplex (1980). Hidden until 60 s of data is in the window.
 - **Personal classifier** — opt-in: hold each emotion for 3 s; a diagonal-Gaussian Naive Bayes fits to your samples and runs alongside the model.
-- **Why? / Ask / Summarize / Discuss recording** — Claude Haiku reads the muscles + V/A + cognitive states + personal templates and explains them in plain language. Multi-turn chat with conversation history.
+- **Why? / Ask / Summarize / Discuss recording** — an LLM reads the muscles + V/A + cognitive states + personal templates and explains them in plain language. Multi-turn chat with conversation history.
 - **Recording** + **JSON export** — clip up to 10 minutes of session data for later analysis or research.
 
 ## Privacy
 
 - Webcam frames are processed locally. **No frames ever leave the device.**
-- The "Why?" feature sends only numerical features (calibrated probabilities, blendshape activations, valence / arousal) to Claude, and only when you click.
+- The "Why?" feature sends only numerical features (calibrated probabilities, blendshape activations, valence / arousal) to the LLM, and only when you click.
 - Personal templates and saved recordings stay in your browser tab. Persistence to `localStorage` is opt-in and off by default; "Clear stored data" wipes it instantly.
 - No analytics, no tracking, no cookies.
 
@@ -54,7 +58,7 @@ Open <http://127.0.0.1:5173/>.
 Two pieces:
 
 1. **Frontend** → Hugging Face Spaces (Static SDK).
-2. **Claude proxy** → Cloudflare Worker (free tier). Holds the Anthropic API key as a server-side secret so the browser bundle never sees it.
+2. **LLM proxy** → Cloudflare Worker (free tier). Holds the API key as a server-side secret so the browser bundle never sees it.
 
 ### 1. Cloudflare Worker (one-time)
 
@@ -89,7 +93,7 @@ git commit -m "Deploy"
 git push
 ```
 
-The Space will auto-build (instant for static) and serve at `https://<your-username>-<space-name>.hf.space`.
+The Space serves at `https://<your-username>-<space-name>.static.hf.space`.
 
 ## Stack
 
@@ -98,7 +102,7 @@ The Space will auto-build (instant for static) and serve at `https://<your-usern
 | Face landmarks + 52 ARKit blendshapes | [`@mediapipe/tasks-vision`](https://www.npmjs.com/package/@mediapipe/tasks-vision) | Apache-2.0 |
 | Emotion classifier | [HSEmotion `enet_b0_8_va_mtl`](https://github.com/HSE-asavchenko/face-emotion-recognition) | Apache-2.0 |
 | ONNX inference | [`onnxruntime-web`](https://www.npmjs.com/package/onnxruntime-web) | MIT |
-| AI explanations | [Claude Haiku 4.5](https://docs.anthropic.com/en/docs/about-claude/models) (via proxy) | API |
+| AI explanations | LLM (proxied via Cloudflare Worker) | API |
 | Build + UI | Vite, TypeScript, Tailwind CDN, Inter / JetBrains Mono | MIT / Apache-2.0 |
 
 ## Research backing the cognitive-states layer

index.html

@@ -6,7 +6,7 @@
     <title>Live Expression Reader</title>
     <meta
       name="description"
-      content="Real-time facial expression analysis with calibrated confidence, cognitive states, and FACS-grounded AI explanations. Runs entirely in your browser."
+      content="Real-time facial expression analysis with calibrated confidence, cognitive states, and FACS-grounded explanations. Runs entirely in your browser."
     />
     <script src="https://cdn.tailwindcss.com"></script>
     <link rel="preconnect" href="https://fonts.googleapis.com" />
@@ -29,7 +29,7 @@
         color: rgb(229, 229, 229);
       }
     </style>
-    <script type="module" crossorigin src="./assets/index-QKvWby_5.js"></script>
+    <script type="module" crossorigin src="./assets/index-Cz6a28xH.js"></script>
   </head>
   <body
     class="bg-gradient-to-b from-neutral-950 via-neutral-950 to-neutral-900 text-neutral-100 min-h-screen"
@@ -321,13 +321,22 @@
       </section>
 
       <footer
-        class="mt-10 pt-4 border-t border-neutral-900 text-xs text-neutral-500 flex flex-wrap gap-x-3 gap-y-1"
+        class="mt-10 pt-4 border-t border-neutral-900 text-xs text-neutral-500 flex flex-wrap gap-x-3 gap-y-1 items-center"
       >
         <span>Apache-2.0 / MIT throughout</span>
         <span aria-hidden="true">·</span>
-        <span>HSEmotion + MediaPipe + Claude Haiku 4.5</span>
+        <span>HSEmotion + MediaPipe + LLM (proxied)</span>
         <span aria-hidden="true">·</span>
         <span>Calibrated, in-browser, no tracking</span>
+        <span aria-hidden="true">·</span>
+        <a
+          href="https://huggingface.co/spaces/arjun10g/live-expression-reader/tree/main"
+          target="_blank"
+          rel="noopener"
+          class="inline-flex items-center gap-1 text-neutral-400 hover:text-indigo-300 underline decoration-dotted underline-offset-2 transition-colors"
+        >
+          View source ↗
+        </a>
       </footer>
     </main>
 
@@ -445,11 +454,11 @@
               </h3>
               <ul class="space-y-1.5 text-neutral-300">
                 <li>
-                  <kbd>Why?</kbd> — Claude summarizes what's happening over
+                  <kbd>Why?</kbd> — the AI summarizes what's happening over
                   the slider window (drag the slider to set 2–30 s).
                 </li>
                 <li>
-                  <kbd>Summarize session</kbd> — Claude summarizes the entire
+                  <kbd>Summarize session</kbd> — the AI summarizes the entire
                   session since you last calibrated.
                 </li>
                 <li>
@@ -471,8 +480,9 @@
               <ul class="space-y-1.5 text-neutral-300">
                 <li>
                   <kbd>Record</kbd> — captures a clip of session data (max
-                  10 minutes). After stopping, click <kbd>Discuss recording</kbd>
-                  to have the AI analyze the whole clip.
+                  10 minutes). After stopping, click
+                  <kbd>Discuss recording</kbd> to have the AI analyze the
+                  whole clip.
                 </li>
                 <li>
                   <kbd>Export</kbd> — downloads the session timeline as a JSON
@@ -523,11 +533,30 @@
               <p class="text-emerald-100/90 text-xs leading-relaxed">
                 Webcam frames stay on this device. Only numerical features
                 (blendshapes, calibrated probabilities, valence / arousal)
-                are sent to Claude, and only when you click Why? / Ask /
+                are sent to the AI, and only when you click Why? / Ask /
                 Summarize. Personal templates and saved recordings live only
                 in this browser tab; the storage opt-in is off by default.
               </p>
             </section>
+
+            <section
+              class="rounded-lg bg-neutral-950/60 border border-neutral-800 p-3"
+            >
+              <h3 class="font-semibold text-neutral-200 mb-1.5">
+                Source &amp; license
+              </h3>
+              <p class="text-neutral-400 text-xs leading-relaxed">
+                Apache-2.0. Built on HSEmotion, MediaPipe, onnxruntime-web,
+                Vite, and TypeScript.
+                <a
+                  href="https://huggingface.co/spaces/arjun10g/live-expression-reader/tree/main"
+                  target="_blank"
+                  rel="noopener"
+                  class="text-indigo-300 hover:text-indigo-200 underline decoration-dotted underline-offset-2"
+                  >View the source on Hugging Face ↗</a
+                >.
+              </p>
+            </section>
           </div>
 
           <div class="mt-6 flex justify-end">