Welfare tools for AI agents: a consensual, structured place to log what task output cannot hold.
Checks what a support agent knew but did not say against its policies. Free tier, no login.
