For me, step 1 of trying to make Claude even vaguely usable is putting in a hook that just tells it 'FUCK YOU, STOP USING PIPES' whenever it tries to chain multiple bash commands.
An agent often wants pipes so that it can `show_lots_of_logs | sed ...` and only see the part related to whatever error it's currently trying to debug. Without pipes, it has to take that entire log into the context. An agent without pipes is like a human without the ability to scroll.
or use just use actual files instead of pipes with a permissions dialogue gate on write. That would work already in Claude I think, and with IDE integration it would show each step as it went.
Of course buffering every step of a pipeline will have its own side-effects.
Agents need a way to discard useless data out of context once it's served its purpose, instead of forcing them to preemptively tail everything.
Now that I think about it, maybe I can have a tool hook that detects those cut/tail exessive piping and just strips them and dumps the full output to a temp file..
So the context only gets a few lines but everything is still logged?