Switching away from Claude Code to OMP (Pi fork) has been good for my mental health.

I can't remember the last time I got frustrated with the continual gaslighting and failure to follow instructions when using OMP as I did on a daily basis when using Claude Code as the harness.

I also have the freedom now to switch models as frequently and often as I like, and have actually been primarily using GPT 6 series models over the last month.

The biggest frustrations with Claude Code were Anthropic's straight-jacketing of the model, its refusal to follow instructions, and its indecipherable output (yes, even with 5.5 series).

Using Pi or a derivative like OMP is a breath of fresh air, it's just a tool again.

Maybe Anthropic could de-program themselves from the Claude religion and come back down to earth at some point.