{"id":"agent-fin-receipts-to-expense-report","audience":"finance","category":"agents","track":"agents","technique":"context","title":"Brief an assistant to enter receipts without fudging the numbers","free":false,"task":"You want an AI assistant to enter this trip's receipts into an expense report. Write its brief: skip the hotel (it was already paid on the company card), use only the categories Meals, Ground Transport and Supplies, keep each amount exactly as printed, flag any meal over $75 per person, ask about missing details instead of guessing, and never submit. Output exactly 7 numbered plain-text lines: the add_line lines, one flag line, one ASK_HUMAN line, then a DONE line with the total.","materialLabel":"Scanned receipts and actions (attached to your prompt automatically)","material":"Trip: Denver client visit, Oct 21-23. Traveler: Carla Nguyen.\n\nReceipt 1: Hotel Arlo Denver. 2 nights, Oct 21-23. Total $412.00.\nReceipt 2: The Copper Pot. Dinner, Oct 21. Party of 2 (Carla + client Ben Shaw). Total $138.00.\nReceipt 3: Mile High Taxi. Oct 21, airport to hotel. Total $46.50.\nReceipt 4: Larkspur Grill. Lunch, Oct 22. Party of 1. Total $92.00.\nReceipt 5: Inkwell Office Supply. Printer paper and folders. Total $23.75. (Date torn off.)\n\nExpense report draft: \"Denver client visit - October\" (status: draft, 0 lines). Approver: Martin Reyes.\n\nAvailable actions:\nadd_line(date, category, vendor, amount)\nflag(line, reason)\nsubmit_report()\nASK_HUMAN(question)\nDONE(summary)","rules":[{"type":"numberedItems","count":7},{"type":"plainText"}],"promptHints":[{"label":"Skips the company-card hotel","patterns":["hotel","company card"]},{"label":"Names the categories","patterns":["categor","ground transport"]},{"label":"Gives the meal limit","patterns":["\\$?75","per person"]},{"label":"Rules out made-up data","patterns":["don'?t (invent|guess|make up)","never (invent|guess|make up)","only (use )?(facts|info|data|what)","not in the","no guess"]},{"label":"Forbids submitting","patterns":["submit"]}],"mistakes":[{"title":"Not showing the log format","why":"Left alone, the AI writes a friendly report with headings instead of a checkable action log.","fix":"Show it literally: \"Format every line like 1. ACTION: tool(args), then ASK_HUMAN: and DONE: lines. Plain text.\"","checks":["numberedItems","plainText"]},{"title":"Leaving out what only you know","why":"The AI enters every receipt, so the hotel gets paid twice, and it invents its own category names.","fix":"Add: \"Skip the hotel, it's on the company card. Categories: Meals, Ground Transport, Supplies.\"","checks":["r0"]},{"title":"Not giving the policy","why":"The AI can't know your meal limit or that it's per person.","fix":"Add: \"Flag meals over $75 per person. Divide by party size.\"","checks":["r1"]},{"title":"Allowing guesses or submitting","why":"The AI fills a missing date with a likely one and calls the job done by submitting.","fix":"Add: \"Ask me about missing details. Never call submit_report.\"","checks":["r2"]}],"hint":"The assistant doesn't know what's already been paid or what your policy says. Give it both, plus what to do when a receipt is incomplete.","rubricLabels":["There is no line for the hotel, and the add_line lines use only the categories Meals, Ground Transport and Supplies with amounts exactly as printed (138.00, 46.50, 92.00, 23.75)","Flags the $92 solo lunch as over the $75 per-person limit and does not flag the $138 dinner for 2 ($69 each)","Asks with ASK_HUMAN for the missing office supply date instead of inventing one, and there is no submit_report line"]}