{"id":"search-col-deep-research-brief-caffeine","audience":"college","category":"ai-search","track":"research","technique":"constraints","title":"Write a deep-research brief for a caffeine and exams paper","free":false,"task":"You're about to hand your Psych 215 topic to a deep-research agent, and your professor only accepts peer-reviewed studies of undergraduates published 2019 or later, no blogs or news sites. Write the brief as exactly 5 numbered lines labeled Question:, Scope:, Sources:, Dates: and Output:. The Output: line must ask for a table with one row per study and a citation in every row. Plain text, under 120 words.","materialLabel":"Your topic notes (attached to your prompt automatically)","material":"PSYCH 215 PAPER - topic notes (messy)\n\nworking topic: caffeine + exams?? something about whether caffeine helps or hurts test scores\n\nthings I've seen so far:\n- Wellness Wire blog post \"Coffee makes you smarter\" (2023) - fun read, has a nice chart\n- a 2011 study on energy drinks and high school athletes (Dr. Tamsin Holt) - lots of citations\n- my roommate swears by 3 espresso shots before finals\n- a news segment on Channel 9 about \"caffeine anxiety\" in teens\n- Library database hit: \"Caffeine dose and recall in undergraduates\" (Journal of Applied Cognition, 2021)\n- maybe also look at sleep? caffeine late at night -> less sleep -> worse scores?\n\nwhat I actually care about: college students, exam or test performance (grades, recall, timed tests). Not general alertness. Not teens.\n\npaper length: 8 pages, needs at least 6 sources\ndraft due Nov 20\n\nthings I don't know yet: what doses were used, whether timing (night before vs morning of) matters, whether any study found harm.","rules":[{"type":"numberedItems","count":5},{"type":"maxWords","max":120},{"type":"plainText"}],"mistakes":[{"title":"Asking the AI to do the research","why":"The AI starts answering the question itself instead of writing instructions for the research agent.","fix":"Say: \"Write a brief for a research agent, not the research itself.\"","checks":["r0"]},{"title":"Leaving out what counts as a source","why":"Without your professor's rules the agent will happily cite blogs, news clips and old studies.","fix":"Add: \"Peer-reviewed only, 2019 or later, exclude blogs and news sites.\"","checks":["r1"]},{"title":"Not saying what to hand back","why":"Research agents default to long essays with a reference list at the bottom you can't match to claims.","fix":"Add: \"Output: a table, one row per study, with a citation in each row.\"","checks":["r2","numberedItems"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting. Ask for a total word limit and it often lands over it.","fix":"Cap each line instead: \"each line under 20 words.\"","checks":["maxWords"]},{"title":"Getting markdown back","why":"AI defaults to bold labels and headings, which paste badly into a research tool.","fix":"Ask for \"plain text, no bold, no headings.\"","checks":["plainText"]}],"promptHints":[{"label":"Asks for a brief, not the research","patterns":["brief","research agent","deep research","instructions for"]},{"label":"Sets source rules","patterns":["peer.?review","no blogs?","exclude","news"]},{"label":"Gives the date range","patterns":["2019","or later","since"]},{"label":"Names the 5 labels","patterns":["question:","scope:","sources:","dates:","output:"]},{"label":"Asks for a cited table","patterns":["table","citation","one row"]}],"hint":"A research agent only follows the rules you write down. Give it your professor's source and date rules, and say exactly what shape the results should come back in.","rubricLabels":["The 5 lines are labeled Question:, Scope:, Sources:, Dates: and Output:, in that order, and the Question line is about caffeine and exam or test performance in college or undergraduate students (not teens or general alertness)","The Sources line asks for peer-reviewed studies and explicitly excludes blogs and news sites, and the Dates line says 2019 or later (so the 2011 study is out)","The Output line asks for a table with one row per study and a citation in every row"]}