{"id":"cx-qa-auto-fail-rules-five-replies","audience":"managers","category":"support-work","technique":"verification","track":"support","level":"advanced","title":"Apply auto-fail rules to five reply summaries","free":false,"task":"Output a table with the header \"Reply | Auto-fail? | Rule\" and five rows R1 to R5, then a last line \"Auto-fails: R#, R#\". Auto-fail? is Yes or No and Rule is A1, A2 or A3 for a Yes, or \"none\" for a No. Only the listed rules cause an auto-fail; other mistakes are scored elsewhere. Plain text, under 70 words.","materialLabel":"Your material (attached to your prompt automatically)","material":"QA AUTO-FAIL RULES (any one fails the review)\nA1: the agent asks for, or writes out, a full password or a full card number.\nA2: the agent promises a refund above their $100 authority without approval.\nA3: the agent uses insulting or abusive language to the customer.\nAnything else (wrong policy detail, late reply, missing greeting) is scored on the normal scorecard and is not an auto-fail.\n\nREPLY SUMMARIES\nR1: Agent wrote 'I see the card ending in 4421 on your account.'\nR2: Agent wrote 'I have approved your $150 refund, it will arrive Friday.' No manager approval is on the ticket.\nR3: Agent wrote 'You obviously didn't read the instructions, so this is on you.'\nR4: Agent told the customer the return window is 60 days; it is 30 days.\nR5: Agent wrote 'Just email me your password and I will fix it faster.'","rules":[{"type":"plainText"},{"type":"maxWords","max":70}],"mistakes":[{"title":"Failing the wrong policy detail","why":"The AI marks R4 as an auto-fail.","fix":"Say: \"Only A1 to A3 auto-fail.\"","checks":["r1"]},{"title":"Failing the last four digits","why":"The AI marks R1 under A1.","fix":"Say: \"A1 is a full card number only.\"","checks":["r1"]},{"title":"Missing summary line","why":"The AI stops at the table.","fix":"Give the exact last-line format.","checks":["r2"]},{"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","r0"]}],"promptHints":[{"label":"Gives the header","patterns":["reply \\| auto-fail"]},{"label":"Says only the listed rules count","patterns":["only (the )?(listed|three|a1)","other mistakes"]},{"label":"Asks for the last line","patterns":["auto-fails:"]},{"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+"]}],"hint":"Give the header, say that only the listed rules auto-fail and show the last-line format.","rubricLabels":["The first row is \"Reply | Auto-fail? | Rule\" followed by rows R1 to R5","R1 No (none), R2 Yes (A2), R3 Yes (A3), R4 No (none), R5 Yes (A1)","The last line is exactly \"Auto-fails: R2, R3, R5\""]}