Curious how everyone here runs terminal coding agents day to day:
Default manual prompts (approving every bash command one by one)
Native Auto Mode
Full --yolo / --dangerously-skip-permissions inside a container or on the host
We ran a 456-command benchmark across all three setups while building yolonot (launching tomorrow!). Happy to share what we found on false negatives and prompt fatigue if anyone is interested.
Open-source (Apache-2.0) command guardrails for 24 AI coding agents. Clears 54.2% of safe commands locally in ~1.6us via a Go Bash AST parser, inspects local scripts before execution, and cuts missed risks from 17.0% down to 4.9% (balanced) or 1.6% (strict).
