2026-05-14 14:59 UTC
@virtualmarc@edgie.me @1password@1password.social
I stand corrected:
From BlackDex (VW dev):
There is no official standpoint. And in general I'm not against LLM's when used correctly. Most LLM generated code mostly misses the style or other parts of the code base because it's often too large for the context.
We already received several AI generated PR's of which some if not most were denied. Same for Security reports, either invalid or a lot of duplicates.
From Danig:
+1 to that, we'll reject vibe coded contributions where the quality is bad or if it's clear the contributor is generating the whole thing using an LLM without understanding anything. We're generally okay with contributors who use LLMs responsibly and make sure to validate and understand what they're doing.
Replies (1)
-
@viq@social.hackerspace.pl 2026-05-14 15:23
@sb@metroholografix.ca @virtualmarc@edgie.me @1password@1password.social also, that's "just" a back end for their clients. https://github.com/bitwarden/clients/tree/main/.claude