My anecdotal experience using Anthropic models in Cursor vs Claude Code:
Cursor makes the effective IQ of these models drop by a double digit # of IQ points
Noticed this while comparing with Open Code and other agentic cli tools too
I suspect Cursor’s system prompts are just bloated context-overloading garbage 🤷♂️
1
0
2
I love this energy
Someone create a prediction market for this please:
“Will @dwr fix the <title> tag placement bug in @vitalik.eth’s open source blog generator”
0
0
2



