Hi @Romain_Mangattale, I am so sorry for the trouble.
Unfortunately I couldn’t find an n8n cloud instance under your forum email, but based on your description I wonder if you might be affected by a bug fixed with n8n version 1.11.1 where the final execution status isn’t written as expected to the database under certain circumstances.
This would lead to the behaviour you have reported where the workflow works and you’re getting the expected response, but the execution still shows up as “running” afterwards.
Can you try upgrading to n8n 1.11.1 via your n8n cloud dashboard and confirm if this problem still happens for newly started executions afterwards?