AAhn·2026-10-19·GuardrailsGate refunds and DB writes behind deterministic checks, not prompt hopesAn Ask HN on controlling agents that take real actions converges on: LLM proposes → deterministic validator (rules, spend ceilings, replay checks) → human approve—especially for refunds, payments, and database writes where prompt 'never do X' fails.AI Agents▲Upvote0I want thisSource ↗Prompt
GBx·2026-10-18·Grok-BotKids draft email with Grok Bot — rule: Don't send itParent workflow: kids ask Grok Bot for a draft email, read it, then show a grown-up first; the standing rule is three words — Don't send it.Grok Bot▲Upvote0I want thisSource ↗Prompt