Find the protocol errors in this generated tool continuation.
Python
def continue_tool(response, messages):
call = response["content"][0]
output = run_tool(call["name"], call["input"])
messages.append({
"role": "user",
"content": [{
"type": "tool_result",
"tool_use_id": call["id"],
"content": output,
}],
})
return messages