Perhaps true. But Sikuli automates the keyboard, mouse, and the screen using OpenCV/Tesseract for object and OCR detection and programming logic.

The only way you'd be able to detect that is by measuring the amount of ms between clicks. And that's defeatable by adding in a random range of 0-300 ms per click. And send the data to a private IRC room. Node-Red can easily handle the data from there.

Now, it looks just like a human.

I think a completely rational company should try to understand the problem. This is a relatively "known" problem with Slack and other realtime chat (i.e. the timesink issue). It's one reason I'm always cautious every time someone at work gets excited about "soon we'll have Slack!!!" based on experiences at my last employer where it sucked up productivity like a black hole.

I've seen a growing number of people actively moving away from Slack or complaining about it due to these issues. How can Slack fix that? Close off an API to stop hiving off users? Well the explicit intent of this approach is to use both tools so it doesn't seem like they lose much.

Either way; I'd argue that a rational company would look at why it's customers were doing these things and then either a) add that feature to the product or b) let things alone.

Slightly different if it's an "export your Slack chat to A.N.Competitor".