I followed the instructions for integrating squeez with Claude Code, i.e. I added the suggested prompt to my CLAUDE.md
However then Claude Code uses it almost never. I made some prompt changes that helped a little bit,
but overall I still see it using it very rarely.
Besides there are also many non-bash tools like Read where it can not be used in principle.
Is there a way to get Claude to use it more?
Maybe with hooks?
There are a lot of competing tools, e.g. https://github.com/rtk-ai/rtk that use hooks to rewrite commands. Maybe that's an option?
Also to increase confidence this actually helps maybe running some benchmarks on swe-bench / swe-rebench would be a good idea.
I followed the instructions for integrating squeez with Claude Code, i.e. I added the suggested prompt to my CLAUDE.md
However then Claude Code uses it almost never. I made some prompt changes that helped a little bit,
but overall I still see it using it very rarely.
Besides there are also many non-bash tools like Read where it can not be used in principle.
Is there a way to get Claude to use it more?
Maybe with hooks?
There are a lot of competing tools, e.g. https://github.com/rtk-ai/rtk that use hooks to rewrite commands. Maybe that's an option?
Also to increase confidence this actually helps maybe running some benchmarks on swe-bench / swe-rebench would be a good idea.