Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by nfg on Wednesday March 18, 2026 @02:28AM
(#47422213)
As a matter of interest are you using the copilot cli?
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by whalesalad on Wednesday March 18, 2026 @09:37AM
(#47425668)
yeah. copilot cli using opus 4.6 vs claude code using opus 4.6
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by nfg on Wednesday March 18, 2026 @04:50PM
(#47431283)
If you could share I’d be really interested in hearing a concrete example of the two behaving differently. I work in Microsoft (not on copilot - though I’m an heavy user, and use Claude code in a personal capacity) and would be quite happy to repro and report back to the copilot cli team who are responsive.
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by chaostheory on Wednesday March 18, 2026 @07:38AM
(#47424409)
I have the same experience with Antigravity and Gemini CLI, both using Gemini 3 Pro. CLI works on the problem with more effort and time. Meanwhile, antigravity writes shitty python scripts for a few seconds and calls it a day. The agent harness matters a lot
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by Atotalnoob on Wednesday March 18, 2026 @09:59AM
(#47425904)
Copilot feels like being a caveman, Claude code feels like modern times comparatively.
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by gtirloni on Wednesday March 18, 2026 @11:07AM
(#47426706)
I think this shows that the model alone isn't the complete story and that these "harnesses" (as people seem to be calling them) shape a lot of the experienced behavior of these tools.
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by theshrike79 on Thursday March 19, 2026 @05:15PM
(#47446237)
My analogy is that the model is the engine and the harness is the driver and chassis.
You can have the biggest monster of an engine ever, but if you put it in a tricycle and a grandma is driving, you won't get good results.
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by codebolt on Wednesday March 18, 2026 @03:53PM
(#47430616)
Opus 4.6 has a 200k context limit in Copilot. Could be the issue.
Re:Get Shit Done: A Meta-Prompting, Context Engineering and ...
by ValentineC on Tuesday March 31, 2026 @04:16AM
(#47584197)
IIRC the context limit in Copilot is actually 128k, not 200k.
200k is the normal context limit elsewhere.
I’ve been really enjoying Codex CLI recently though. It seems to do just as well as Opus 4.6, but using the standard GPT 5.4