just out of curiosity why does harness take longer...
# proj-agent-canvas
f
just out of curiosity why does harness take longer to run that say usual harnesses (opencode etc)
e
Thanks for the feedback! Can you say more? Do you mean when having an agent conversation that turns takes longer? Curious if you're noticing this when using the same model on Canvas vs other harnesses.
l
Yeah, and if you could share what LM you’re using that’d also be helpful
f
tauri apps for different trading desks
I am trying to use the openhands harness - inside another harness
we use portkey at our firm - so it is hard with openhands - so i get say claude code to run it in the openhands harnes
👀 1
let grab some more detials
i
I've heard other users reporting issues with Portkey
🙌 1
f
cheers yer im not a fan tbh
so now im using github copilot harness - what is the easiest way to use openhands
headless?
a
The system prompt for openhands emphasizes completeness and thoroughness. So it will often do a lot more work/take longer than a harness that doesn't push for that.
h
> so now im using github copilot harness - what is the easiest way to use openhands Is agent-canvas not an option? Either with npm/npx or even the electron app
f
apparently im using agent canvas through mcp (?)