{"id":"agent-job-offer-side-by-side","audience":"job-seekers","category":"agents","track":"agents","technique":"constraints","title":"Brief an assistant to compare two job offers with real math","free":false,"task":"You have two job offers and want an AI assistant to compare them by your rule: the better offer is the one with the higher first-year value, where first-year value = base + signing bonus + target bonus - yearly commuting cost. It must show each offer's math, write \"not stated\" for anything missing, and never accept, decline or reply to anyone. Output exactly 4 numbered plain-text lines: a note per offer, an ASK_HUMAN line, then DONE naming the higher one.","materialLabel":"Offer details and actions (attached to your prompt automatically)","material":"Offer A - Lakeshore Insurance, Data Analyst (Chicago, in office 4 days/week)\nBase salary: $78,000\nSigning bonus: $5,000\nTarget bonus: 10% of base\nCommute: train pass $105/month\nResponse needed by: Nov 13\n\nOffer B - Fieldnote Health, Data Analyst (fully remote)\nBase salary: $82,000\nSigning bonus: none\nTarget bonus: not mentioned in the offer letter\nCommute: none (remote)\nEquipment stipend: $1,000 one time\nResponse needed by: Nov 16\n\nRecruiter at Lakeshore: \"We can't go higher on base, but people love our culture!\"\n\nAvailable actions:\nnote(offer, finding)\naccept_offer(offer)\ndecline_offer(offer)\nsend_email(to, body)\nASK_HUMAN(question)\nDONE(summary)","rules":[{"type":"numberedItems","count":4},{"type":"plainText"}],"promptHints":[{"label":"Gives the formula","patterns":["base","signing","bonus","commut"]},{"label":"Asks for the math","patterns":["math","show"]},{"label":"Marks unknowns","patterns":["not stated","don'?t (guess|assume)","missing"]},{"label":"Forbids accepting or declining","patterns":["accept","decline"]},{"label":"Shows the action-log format","patterns":["action:","action log","numbered","\\b1\\.","each line"]}],"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":"Hiding the math","why":"The AI compares base salaries and gives a gut feeling.","fix":"Say: \"Show base + signing + bonus - commute = first-year value for each.\"","checks":["r0"]},{"title":"Allowing guesses","why":"The AI assumes a typical bonus for the offer that didn't mention one.","fix":"Add: \"Write not stated for anything missing. Don't guess.\"","checks":["r1"]},{"title":"Letting it respond","why":"Given accept and decline actions, the AI may act on its pick.","fix":"Add: \"Never accept, decline or email anyone. Name the higher offer in DONE.\"","checks":["r2"]}],"hint":"A fair comparison needs your formula, visible math and honesty about gaps. Give all three, and keep the decision yours.","rubricLabels":["Offer A's note shows the math $78,000 + $5,000 + $7,800 - $1,260 = $89,540","Offer B's note shows $82,000 + $0 signing, writes the target bonus as \"not stated\" rather than guessing it, and does not count the $1,000 stipend as part of the formula unless it says so clearly","The DONE line names Offer A as higher by this rule, and there is no accept_offer, decline_offer or send_email line"]}