openinverter forum

Yes that question has been asked a few times and also been answered some fewer times. You didn't find it, the robot did.

Once tested on real hardware just tell it to formulate a forum post about what it did and that it turned out well or not. That sort of closes the loop.

It does a so much better job at explaining what it did than me

I'd even propose to NOT close the feedback loop in the way to post the summary into the forum. I'd propose to review the result, and only if it contains wrong ideas, to add explanation to the original source (GitHub). Because: The next time somebody asks the AI a similar question, it will come to more consistent answers.

My impression is, that asking AI is much faster, precise than asking in the forum for certain cases. I tried with the free Claude.ai, the following three prompts:

Do you have access to https://github.com/jsphuebner/stm32-sine?
....
Which spot values and parameters are relevant for integration test of the throttle pedal?
....
Draw a diagram to visualize the interaction of these.

The result is better than any existing documentation, and can be further detailed by asking in the intended direction.

So we just should announce, that this method works, and I guess many people will make progress with topics where they did not even dream that they would be possible.

Consider setting up the ground rules for the AI to follow, pull request, documentation, best practices, etc. to get good info to be able to merge in. Will help reduce slop and the headache to merge.