Update app.py

app.py

@@ -302,6 +302,55 @@ async def health_check():
         "device_count": torch.cuda.device_count() if torch.cuda.is_available() else 0
     }
 
+# Anthropic Claude 格式的请求体
+class MessagesRequest(BaseModel):
+    model: str
+    messages: List[Message]
+    max_tokens: Optional[int] = 2048
+    temperature: Optional[float] = 0.7
+    top_p: Optional[float] = 0.9
+
+@app.post("/v1/messages")
+async def messages_endpoint(request: MessagesRequest):
+    """
+    兼容 Anthropic Claude CLI 的接口
+    """
+    try:
+        if model is None or tokenizer is None:
+            try:
+                load_model()
+            except:
+                raise HTTPException(status_code=503, detail="Model not loaded")
+
+        # 使用已有的格式化和生成逻辑
+        prompt = format_messages(request.messages)
+        response_text = generate_response(
+            prompt,
+            request.temperature,
+            request.max_tokens,
+            request.top_p
+        )
+
+        return {
+            "id": f"msg-{uuid.uuid4().hex[:8]}",
+            "type": "message",
+            "role": "assistant",
+            "content": [
+                {"type": "text", "text": response_text}
+            ],
+            "model": request.model,
+            "stop_reason": "end_turn",
+            "stop_sequence": None,
+            "usage": {
+                "input_tokens": len(tokenizer.encode(prompt)),
+                "output_tokens": len(tokenizer.encode(response_text))
+            }
+        }
+
+    except Exception as e:
+        print(f"Error processing /v1/messages request: {str(e)}")
+        raise HTTPException(status_code=500, detail=f"Internal server error: {str(e)}")
+
 if __name__ == "__main__":
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=7860)