What each part of the prompt does
- Five-line summary first
- Question, method, finding, limitation, relevance. Those five lines are what a tutor asks about in a seminar, and writing them first forces the model to commit to what the paper is actually claiming before it starts listing details.
- Purpose as an input
- [PURPOSE] changes what counts as important. Notes for an essay source need the limitation and the exact claim; notes for an exam need the method and definitions. Saying which one you need shapes every later step.
- Numbers with their location
- Exact figures tagged with their section make the notes trustworthy. When you quote a sample size in an essay, you can go straight back to the right part of the paper and check it.
- Source-only rule
- Claude knows a lot about most fields, which is a risk here. The rule to use only the pasted text, and to write "check in the paper" when unsure, keeps outside facts from slipping into notes you will later treat as the paper's own.
Variations to try
Same idea, one thing changed.
Comparing two papers <task>Turn the research paper below into study notes I can revise from and an outline I can reuse in my own writing.</task> <context> - My course and level: [COURSE] - Why I am reading this paper: [PURPOSE] - Terms I already know and do not need explained: [KNOWN TERMS] </context> <paper> [PAPER TEXT] </paper> I will paste a second paper on the same question next. After both, add a table comparing their methods, samples and findings, and say where they disagree. <instructions> 1. Start with a five-line summary: the question the paper asks, the method, the main finding, the biggest limitation, and why it matters for [PURPOSE]. 2. Build a section-by-section outline that follows the paper's own structure. Under each heading list the key points as short bullets in my words, not copied sentences. 3. Add a "Terms" list for every technical term not in [KNOWN TERMS], each with a one-line definition based on how this paper uses it. 4. Pull out the numbers that carry the argument (sample sizes, effect sizes, dates) exactly as the paper states them, with the section they come from. 5. Mark anything you are unsure about as "check in the paper" instead of guessing. 6. End with five recall questions that cover the method and the findings, answers hidden under a separate heading. </instructions> <rules>Use only the text I provide. Do not add findings, citations or numbers from elsewhere.</rules>
Notes for a seminar presentation <task>Turn the research paper below into study notes I can revise from and an outline I can reuse in my own writing.</task> <context> - My course and level: [COURSE] - Why I am reading this paper: [PURPOSE] - Terms I already know and do not need explained: [KNOWN TERMS] </context> <paper> [PAPER TEXT] </paper> <instructions> 1. Start with a five-line summary: the question the paper asks, the method, the main finding, the biggest limitation, and why it matters for [PURPOSE]. 2. Build a section-by-section outline that follows the paper's own structure. Under each heading list the key points as short bullets in my words, not copied sentences. 3. Add a "Terms" list for every technical term not in [KNOWN TERMS], each with a one-line definition based on how this paper uses it. 4. Pull out the numbers that carry the argument (sample sizes, effect sizes, dates) exactly as the paper states them, with the section they come from. 5. Mark anything you are unsure about as "check in the paper" instead of guessing. 6. End with three discussion questions I could put to the seminar group, each tied to a specific section of the paper. </instructions> <rules>Use only the text I provide. Do not add findings, citations or numbers from elsewhere.</rules>
If the result isn't right
- For papers longer than your chat allows, paste the abstract, methods and results first; the discussion section is the easiest part to summarise later.
- Reread the limitation line before citing the paper in an essay; markers notice when a source is used to support more than it found.
- If a figure or table carries the result, describe it in a sentence when you paste the text, because pasted PDFs often drop them.