Update app.py

app.py

@@ -1,18 +1,21 @@
 from flask import Flask, request, jsonify
 from flask_cors import CORS
 import os
-import requests
+from huggingface_hub import InferenceClient
 
 app = Flask(__name__)
-CORS(app)  # Bypasses your browser security blocks instantly
+CORS(app)  # Keeps your Acode connection unlocked permanently!
 
+# Pulls your hidden secret token safely
 HF_TOKEN = os.getenv("HF_TOKEN")
 MODEL_REPO = "Kelvin000010191/Krypton-1"
-API_URL = f"https://api-inference.huggingface.co/models/{MODEL_REPO}"
+
+# Initializes the official native Hugging Face client
+client = InferenceClient(model=MODEL_REPO, token=HF_TOKEN)
 
 @app.route("/", methods=["GET"])
 def home():
-    return "Krypton-1 Core API Proxy Node Online."
+    return "Krypton-1 Native Proxy Node is Online."
 
 @app.route("/api/chat", methods=["POST"])
 def chat():
@@ -20,32 +23,24 @@ def chat():
         data = request.json
         user_prompt = data.get("prompt", "")
         
-        headers = {
-            "Authorization": f"Bearer {HF_TOKEN}",
-            "Content-Type": "application/json"
-        }
+        # Uses internal channel loops instead of outbound internet requests
+        response = client.text_generation(
+            user_prompt,
+            max_new_tokens=250,
+            temperature=0.7,
+            return_full_text=False
+        )
         
-        payload = {
-            "inputs": user_prompt,
-            "parameters": {
-                "max_new_tokens": 250,
-                "temperature": 0.7,
-                "return_full_text": False
-            }
-        }
+        return jsonify({"reply": response})
         
-        response = requests.post(API_URL, headers=headers, json=payload)
-        res_data = response.json()
+    except Exception as e:
+        error_msg = str(e)
         
-        # Keep track if your 15GB model needs a hot second to load weights
-        if isinstance(res_data, dict) and "error" in res_data:
-            return jsonify({"reply": f"SYS_BOOT: Krypton-1 is loading weights. Estimated activation: {round(res_data.get('estimated_time', 60))}s. Wait a moment and hit RUN again."})
+        # Catch if your 15GB model cluster is waking up/loading weights
+        if "loading" in error_msg.lower():
+            return jsonify({"reply": "SYS_BOOT: Krypton-1 is loading weights into memory clusters. Let the cloud cook for a moment, then press RUN again."})
             
-        reply = res_data[0].get("generated_text", "Data stream loop empty.")
-        return jsonify({"reply": reply})
-        
-    except Exception as e:
-        return jsonify({"reply": f"SYS_ERR: Cloud proxy connection broken. {str(e)}"})
+        return jsonify({"reply": f"SYS_ERR: Native cloud connection broken. {error_msg}"})
 
 if __name__ == "__main__":
     app.run(host="0.0.0.0", port=7860)