Update app.py

app.py

@@ -17,22 +17,17 @@ subprocess.Popen(["ollama", "serve"])
 
 @app.api_route("/{path:path}", methods=["GET", "POST", "PUT", "DELETE"])
 async def proxy_to_ollama(request: Request, path: str):
-    # Reverse Proxy: Forwards Claude CLI requests to the internal Ollama engine
-    url = httpx.URL(f"http://127.0.0.1:11434/{path}", query=request.url.query.encode("utf-8"))
+    # This captures the '/v1/messages' and sends it to Ollama
+    # Ensure no extra '/v1' is added by the CLI
+    target_path = path if not path.startswith("v1/v1") else path.replace("v1/v1", "v1")
+    url = httpx.URL(f"http://127.0.0.1:11434/{target_path}", query=request.url.query.encode("utf-8"))
     
     async with httpx.AsyncClient(timeout=None) as client:
         req = client.build_request(
-            request.method,
-            url,
-            headers=request.headers.raw,
-            content=request.stream(),
+            request.method, url, headers=request.headers.raw, content=request.stream()
         )
         resp = await client.send(req, stream=True)
-        return StreamingResponse(
-            resp.aiter_raw(),
-            status_code=resp.status_code,
-            headers=dict(resp.headers)
-        )
+        return StreamingResponse(resp.aiter_raw(), status_code=resp.status_code, headers=dict(resp.headers))
 
 @app.get("/")
 def health_check():