Skip to content
X · @teortaxesTex · X / Twitter

Well, maybe this goes to show that it's time for the old harness to retire. I don't see how gimping the agent's long-term memory allows a valid compar…

Well, maybe this goes to show that it's time for the old harness to retire. I don't see how gimping the agent's long-term memory allows a valid comparison to human efficiency.Greg Kamradt: We use the same standard harness for Anthropic/OpusSame completions style endpoint - no reasoning logs passed backThe point is to b