*bare
I built a policy authorizer on top of pi-permission-system that looks kinda like nftables that models commands/subcommands/options as read/write and enfores policies depending on if the sub agent is read only or write, which really cuts down on the stupid decisions and nicely complements AI cmd review. Pi shines for making things like that possible.
Wrap your agent of choice in e.g. docker sandbox, nono.sh, Nvidia OpenShell, microsanbox or any other sandboxing tool and run it without any checks (i.e. "yolo mode", dangerously-skip-permissions etc.)
IMHO it makes the whole category of problems around permission fatigue, trusting the harness (or plugin) authors etc go away.
Strong sandbox + disabling permission checks gives me maximum capabilities, least annoyance and (relative) peace of mind that any prompt injection, malicious plugin,... won't wreck my shit or leak credentials of any kind
I gave the harness a try and it is refreshing, but the insistence on staying bear bones and not implementing basic functionality natively, together with no vetting process for popular 3rd party modules, made me return to Claude Code.