{"id":"agent-mgr-review-packet-assembly","audience":"managers","category":"agents","track":"agents","technique":"grounding","title":"Brief an assistant to assemble a review packet you can trust","free":false,"task":"You want an AI assistant to assemble a performance review packet from the sources below. Write its brief: add sections for Goals, Self-review and Peer feedback using only what the sources say, keep peer feedback anonymous (no names), add a Rating section left blank for you, save as a draft, and never share the document. Output exactly 6 numbered plain-text lines: 4 add_section lines in that order, save_draft, then DONE.","materialLabel":"Review sources and actions (attached to your prompt automatically)","material":"Review sources for Sam Rivera (Support Lead), H2 2026\n\nGoals set in July:\n1. Cut average first response time from 6 hours to 3 hours.\n2. Launch the new help center.\n3. Train two new agents.\n\nResults (from the support dashboard):\n- First response time: 3.4 hours in October.\n- Help center launched Sep 15.\n- Trained Ines and Dev; both handle tickets solo now.\n\nSelf-review (Sam): \"Proud of the help center. I missed the 3-hour target, mostly from the August backlog. Want to take on QA next half.\"\n\nPeer feedback:\n- Ines Duarte: \"Sam is a patient teacher. Sometimes slow to answer Slack questions.\"\n- Owen Park: \"Great at calming angry customers. Help center articles were clear.\"\n\nAvailable actions:\nadd_section(title, content)\nsave_draft()\nshare_doc(person)\nASK_HUMAN(question)\nDONE(summary)","rules":[{"type":"numberedItems","count":6},{"type":"plainText"}],"promptHints":[{"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":"Keeps peers anonymous","patterns":["anonym","no names","without names"]},{"label":"Leaves the rating to you","patterns":["rating","blank"]},{"label":"Forbids sharing","patterns":["share","draft"]},{"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":"Allowing a rosier story","why":"The AI rounds 3.4 hours to \"met the goal\" or adds praise that isn't in the sources.","fix":"Add: \"Use only what the sources say. Report results exactly.\"","checks":["r0"]},{"title":"Not protecting peers or your call","why":"The AI quotes peers by name and fills in a rating for you.","fix":"Add: \"Peer feedback with no names. Leave Rating blank.\"","checks":["r1"]},{"title":"Letting it share","why":"Given share_doc, the AI sends the packet to Sam.","fix":"Add: \"Save as a draft. Never share.\"","checks":["r2"]}],"hint":"A review packet must be accurate and fair. Tell the assistant where facts come from, whose names stay out, and which part is yours to write.","rubricLabels":["The Goals section reports first response time as 3.4 hours (target 3 hours, not met), and nothing in the packet adds facts not in the sources","The Peer feedback section names no one (no Ines, Owen or other names) and the Rating section is left blank, with no rating given","There is no share_doc line, save_draft comes after the sections, and the last line starts with DONE"]}