{"slug": "cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out", "title": "Cap refunds per order, not per action, when AI agents can pay out", "summary": "A developer writing under the AgentRisk project argues that teams deploying AI agents with refund authority should cap total refunds per order rather than per action, since multiple agents or repeated runs on the same order can each pass a per-action check while collectively paying out far more than intended. The writeup recommends a single documented total refund cap per order that every automated and human refund path can see, counting store credits, gift cards and shipping refunds toward the total.", "body_md": "Most teams that let an AI agent issue refunds start with a simple guardrail: a per-action limit. Each refund the agent issues has to stay under some amount, and anything bigger goes to a person.\n\nThat limit is useful, but it answers the wrong question. It tells you how much a single refund can be. It doesn't tell you how much a single order can lose.\n\nPicture an order that's been touched by more than one automated path. A support agent handles the chat. A returns agent processes the return label. A billing agent cleans up a duplicate charge. Each one stays inside its own limit, and each one decides a refund is reasonable.\n\nThree refunds of $100 on the same order is still $300 out the door, and no single rule was broken.\n\nThe same thing happens with one agent, too. If a customer opens three chats about the same order, or a retry runs after a timeout, each run sees a fresh request and applies the same per-action check. Nothing in that check knows what already happened.\n\nThe fix is to make the limit about the order, not the action:\n\nStore credits, gift cards, and shipping refunds count too. If they're tracked separately, the total is wrong.\n\nYou don't need a new system to start. Write down your total refund cap per order, in one sentence, and check whether every refund path, automated or human, can actually see it. If the answer is \"only some of them,\" that's the gap to close first.\n\nIf you run agents that can move money, I'm curious how you've handled this:\n\n*Disclosure: I run AgentRisk, a written risk report for customer-facing AI agents. This post is adapted from one of the short risk tips I publish. More at [https://rleffke.com/agentrisk](https://rleffke.com/agentrisk)*", "url": "https://wpnews.pro/news/cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out", "canonical_source": "https://dev.to/rick_leffke_6541e28129e2a/cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out-3hd", "published_at": "2026-11-13 15:15:14+00:00", "updated_at": "2026-11-13 15:23:40.387449+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "ai-tools"], "entities": ["AgentRisk"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out", "markdown": "https://wpnews.pro/news/cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out.md", "text": "https://wpnews.pro/news/cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out.txt", "jsonld": "https://wpnews.pro/news/cap-refunds-per-order-not-per-action-when-ai-agents-can-pay-out.jsonld"}}