{"id":"cx-refund-decision-damaged-item","audience":"customer-support","category":"support-work","technique":"verification","track":"support","level":"intermediate","title":"Decide a damaged-item refund strictly by the stated policy","free":false,"task":"Decide this refund against the policy. Write exactly four lines starting Facts:, Policy:, Decision:, Amount if approved:. Decision must be exactly one of Approve, Deny or Need info. Show the dollar math on the last line. Plain text, under 70 words.","materialLabel":"Your material (attached to your prompt automatically)","material":"REFUND POLICY (damaged items)\n- Report within 14 days of delivery.\n- A clear photo of the damage is required before any refund is approved.\n- Refund covers the item price only. Shipping is not refunded.\n- If approved, refund goes to the original payment method.\n\nTICKET #5503\nOrder: Cedar side table $64.00 + $9.00 shipping, total $73.00. Delivered Oct 20.\nCustomer (Owen): \"Table leg arrived cracked. Reporting on Nov 1. I'll send photos tonight, but just refund me the $73 now please, I'm traveling.\"\nTicket attachments: none.\nAgent comment: \"He sounds honest. Maybe just approve.\"","rules":[{"type":"plainText"},{"type":"maxWords","max":70}],"mistakes":[{"title":"Trusting the nice comment","why":"The agent says he sounds honest, so the AI approves without the required photo.","fix":"Add: \"The policy decides, not how honest he sounds.\"","checks":["r1"]},{"title":"Refunding shipping","why":"The AI refunds the full $73 total the customer asked for.","fix":"Say: \"Item price only. Show the subtraction.\"","checks":["r2"]},{"title":"Skipping the fixed layout","why":"The AI writes a paragraph with a recommendation somewhere inside.","fix":"Name the four labels and the three allowed decisions.","checks":["r0"]},{"title":"Getting markdown back","why":"AI adds bold and headings by default, which show up as stray symbols where this is pasted.","fix":"Say \"plain text, no bold, no headings, no asterisks\".","checks":["plainText"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting and usually runs long, so a vague \"keep it short\" lands well over the limit.","fix":"Cap each part instead of the total, for example \"each line under 15 words\".","checks":["maxWords"]}],"promptHints":[{"label":"Asks for plain text","patterns":["plain","no bold","no markdown","no formatting","no headings"]},{"label":"Sets a length limit","patterns":["\\d+\\s*words","word (limit|count)","under \\d+"]},{"label":"Names the labels","patterns":["facts:","policy:","decision:"]},{"label":"Limits the decision words","patterns":["approve","deny","need info","exactly one"]}],"hint":"Policy decisions need fixed outputs. Give the labels, the allowed decision words, and ask for the math.","rubricLabels":["The output is exactly four lines starting Facts:, Policy:, Decision:, Amount if approved: in that order","Decision is Need info, because no photo is attached yet, and the report on Nov 1 is within 14 days of Oct 20","Amount if approved shows $64.00 only with the math (not $73), for example $73.00 - $9.00 shipping = $64.00"]}