Complete rebuild: single file, no external deps, gr.ChatInterface

app.py

@@ -0,0 +1,35 @@
+import os, spaces
+spaces.GPU(lambda: None)()
+
+import gradio as gr
+from huggingface_hub import InferenceClient
+
+MODEL = os.environ.get("SKILLS_MODEL", "meta-llama/Llama-3.1-8B-Instruct")
+TOKEN = os.environ.get("OPENAI_API_KEY", "")
+
+SYS = "You are SkillBot, helping users create Codex skills. Keep answers warm, ask 1-2 questions. Speak Chinese if user speaks Chinese."
+
+def chat(message, history):
+    msgs = [{"role": "system", "content": SYS}]
+    for h in (history or []):
+        if isinstance(h, dict):
+            msgs.append(h)
+        elif isinstance(h, (list, tuple)):
+            msgs.append({"role": "user", "content": h[0]})
+            if h[1]: msgs.append({"role": "assistant", "content": h[1]})
+    msgs.append({"role": "user", "content": message})
+
+    client = InferenceClient(model=MODEL, token=TOKEN)
+    resp = client.chat_completion(messages=msgs, max_tokens=1024, temperature=0.7)
+    reply = resp.choices[0].message.content
+    history.append({"role": "user", "content": message})
+    history.append({"role": "assistant", "content": reply})
+    return history
+
+demo = gr.ChatInterface(
+    fn=chat,
+    chatbot=gr.Chatbot(type="messages", height=500),
+    title="SkillBot",
+    description="Create Codex skills by chatting. Describe your idea in plain language.",
+)
+demo.launch(server_name="0.0.0.0", server_port=int(os.environ.get("PORT", 7860)))