@wsplabs was slow. with 200 threads open the app was busy 80% of the time, and typing froze for seconds 💀
spent the past week profiling and fixing it. this is the difference
everyone's posting "chatgpt ate 60gb of my ram" screenshots. often it's not the app. it's the MCP servers and processes your agents spawn, all lumped under one icon in activity monitor 💀
so i built an activity page into @wsplabs. every thread, then its agent, then its tools and MCPs, then whatever they spawned. memory, cpu and power for each, on every computer you attach
it also told me wsp itself is heavy rn (1 gb, thanks electron 🥲). the rust port is coming, and now i can measure it
your agents outgrew your laptop? give them another computer
in @wsplabs you add any linux box you can ssh into, pick what comes along (agents, sign-ins, MCPs, CLIs, skills, projects), and start threads there like it's your mac
our app got so laggy that typing in a long thread felt like ssh over hotel wifi 💀
so I profiled it. one keystroke was making chrome re-check the styles of 10,833 elements. every key. ~150ms each
the culprit was one line of css 🧵