Comment by ivape

This is an interesting way to do micro context engineering. This is basically pulling in the minimum relevant code for your current concern, and then you can just sprinkle on some instructions and send off your prompt. Might work reasonably well for very small local models, or just generally cheap inferencing on the server.

I love this plugin btw.