Post #2626306
2026-05-04 19:20 UTC
@leo@60228.dev @jacksonchen666@micro.jacksonchen666.com oh! that's a useful piece of the puzzle, thanks <3
and yeah, I haven't actually used vscode or anything to test this, but agreed that it must not be triggering on every commit otherwise 4000 is *way* too low. it would be really useful to know what the unclear scenarios that cause the counter to be incremented are, but I'm not gonna buy a copilot subscription to find out. I saw some evidence yesterday suggesting that possibly the thing that was causing people to say it was triggering on non-copilot changes was that it triggered on autocomplete and their mental model of "using copilot" did not include copilot autocomplete ([this issue](https://github.com/microsoft/vscode/issues/313064)). If this is true, then the copilot users are just confused about what "using copilot" means and it isn't really a false positive for detection tools at all.
Replies (1)
-
@olivia@pl.firechicken.net 2026-05-04 19:28
@leo@60228.dev @jacksonchen666@micro.jacksonchen666.com I did also make a second plot yesterday of copilot commits as a proportion of total commits, which is additional evidence to rule out the possibility that the drop after 1.118 was just caused by new commits not having been pushed yet