How is this different than OP CLI with a service account?

Much safer as the technique ensures the AI agent never even sees the creds vs the CLI the inference/context can see them which opens you up to a whole class of security risk.

https://support.1password.com/1password-claude-security/

	▲	laul_pogan 5 hours ago \| parent [-]
		Oh wow! I had no idea. Even with a service worker? I'll switch over to this- it works just as well for insertion and .env replacements in claude code?