in my experience claude’s “research” mode isn’t particularly helpful. i found its reports consistently less useful than normal conversations. haven’t tried comparing to claude code with intermediate artifacts yet. does this track? has anyone had much better experience with research mode?
this from [@replay-io.bsky.social@bsky.brid.gy]( ) seems to be in that vein [blog.replay.io/launching-nu...]( ) didn't see it before, wondering if anyone tried? RE: [Launching Nut.new]( ) View quoted note →
debbie has been doing fantastic work in the playwright ecosystem RE:
liquid glass so far fells unconvincing to me because in stock apps it mostly targets the shell. so it’s like the shell got a new design language ok but what about stuff inside the apps?
ohh i see RE:
found a ViewTransition crash in iOS safari :( [github.com/facebook/rea...]( )
uhhh. what https://morel.us-east.host.bsky.network/xrpc/com.atproto.sync.getBlob?did=did:plc:fpruhuo22xkm5o7ttr2ktxdo&cid=bafkreibdn6ph6tnpqcbmr45eg5sh6hg2sb4w37dlenlqdzlqb4j3i33zuq
i’m finally watching paul mccartney in space and it’s good
should’ve called them API components RE: View quoted note →
llms enabling new modes of perf debugging. just figured out an annoying safari unexpected relayout by passing traces to claude which calculated events/layouts ratio to verify theories the more structured data it can access the better the results