In the call today someone had mentioned the frustration with getting a screenshot from the browser. I noticed it is successful with a python script. I don't recall exactly - mostly because the agent figured it out on it's own. When it figured out how to complete the task with the python script, I told it to save the steps it used as a new skill. Initially it took about 4 minutes to complete the task, but now it only takes about 23 seconds. I am going to make skills for other troublesome browser tasks as well. Has this been the same experience for you?