Invoking a Canvas-generated endpoint via Python script in my desktop

I generated a job in Canvas via CSV file. After deploying, an endpoint was generated.

Now I am trying to generate predictions via a Python script on my desktop. The script is as follows:

import requests
import pandas as pd
import json
import os

# Definir o endpoint do Canvas
endpoint = "https://runtime.sagemaker.us-east-1.amazonaws.com/endpoints/canvas-new-deployment-03-12-2024-4-51-PM/invocations"

# Carregar o arquivo CSV
data = pd.read_csv("C:\\Users\\Druida\\Desktop\\dados_binance.csv")

# Converter o DataFrame para JSON
data_json = data.to_json(orient="records")

# Obter o token de autenticação
token = os.environ.get("canvas-new-deployment-03-12-2024-4-51-PM")

# Criar a requisição
headers = {
    "Content-Type": "application/json",
    "X-Amzn-SageMaker-Custom-Auth": token
}
body = data_json
response = requests.post(endpoint, headers=headers, data=body)

# Tratar a resposta
if response.status_code == 200:
    prediction = json.loads(response.content)
    print(prediction)
else:
    print(f"Erro: {response.status_code}")
    print(response.content)

However, instead of the prediction, I get the following message regarding the token:

C:\Users\Druida\PycharmProjects\pythonProject.venv\Scripts\python.exe "C:\Users\Druida\PycharmProjects\pythonProject\regressão linear.py" Erro: 403 b'{"message":"Missing Authentication Token"}'

Process finished with exit code 0

Please help me because I can't generate the predictions.