5 Claude Prompts for Research Papers (2026)
A five-step research paper workflow that uses Claude as a long-context research assistant — helping you narrow a topic to a focused research question, build a literature review structure, draft academic sections, and produce a properly annotated bibliography without fabricated citations.
Run = copies the prompt and opens the tool. Remix = save an editable copy to your workspace (free account).
Narrow your research question
🟠 Claude
I'm writing a research paper for [COURSE/LEVEL] on the broad topic of: [TOPIC] Length: [PAGE/WORD COUNT] Deadline: [DATE] Help me develop a focused, researchable question: 1. What's the difference between a topic and a research question? Show me using my topic as an example. 2. Generate 5 possible research questions from this topic, ranging from too-broad to too-narrow — and show me why each one is or isn't workable at my paper length. 3. For the 2 most promising questions, tell me: (a) what type of research methodology it implies, (b) what the main bodies of literature I'd need to engage with are, (c) whether it's been thoroughly covered already or has room for a novel contribution 4. Recommend the strongest question and explain why it's the right scope for my assignment.
Why this works: Research papers fail most often at this stage — not execution. A too-broad question produces a shallow survey; a too-narrow question runs out of material. Getting this right at the start saves weeks.
Good output: 5 candidate questions with breadth analysis, methodology implications and literature scope for the top 2, and a recommended question with reasoning.
Structure a literature review
🟠 Claude
My research question is: [CHOSEN QUESTION] Here are the sources I've found so far (paste titles/abstracts, or describe them): [PASTE SOURCES] Help me structure a literature review: 1. Group my sources into 3–4 thematic clusters (not chronological — by argument or approach) 2. For each cluster, identify: the main claim shared by these sources, how they differ from each other, and what they collectively leave unanswered 3. Show me how the clusters connect — what's the narrative arc of my literature review? 4. Flag any important perspective or body of work that seems missing from my sources 5. Draft the opening paragraph of my literature review that explains how the existing research has set up the gap my paper will address
Why this works: Organizing literature thematically (by argument) rather than chronologically (by date) is what separates a literature review from an annotated list of summaries.
Good output: 3–4 thematic clusters with internal similarities/differences, a narrative arc connecting them, gap flags, and a draft opening paragraph.
Draft the methodology or analytical framework
🟠 Claude
My research question: [QUESTION] My paper's approach: [EMPIRICAL (data/experiments) / THEORETICAL / ANALYTICAL / LITERATURE-BASED] Draft the methodology or analytical framework section of my paper (target: [WORD COUNT] words). This section should: - Explain what approach I'm using and why it's appropriate for this research question - Define the key concepts or theoretical framework I'll apply - Describe any limitations of this approach and acknowledge them honestly - Use passive voice where appropriate for academic convention in this field: [YOUR FIELD] - Include [CITATION PLACEHOLDER: Author, Year] wherever I need a source — I'll fill these in from my own research Do NOT invent specific data, studies, or statistics. Use placeholders wherever empirical claims would be needed.
Why this works: The 'citation placeholder' rule is critical — it prevents hallucinated methodological citations, which are among the most damaging errors in academic work.
Good output: A draft methodology section with your approach, framework, limitations — and CITATION PLACEHOLDER tags instead of invented sources.
Draft the analysis and discussion sections
🟠 Claude
Here are my key findings or the main argument of my paper:
[PASTE YOUR NOTES / EVIDENCE / KEY POINTS]
Draft the analysis/discussion section (target: [WORD COUNT] words) that:
- Interprets what these findings mean in relation to my research question
- Connects back to the literature from Step 2: which sources does this confirm, challenge, or complicate?
- Acknowledges limitations and alternative interpretations
- Uses hedged academic language appropriately ('suggests', 'indicates', 'may be' — not 'proves' or 'shows definitively' unless the evidence genuinely supports it)
Use [CITATION PLACEHOLDER] for any source reference. Do not make up data. If I haven't given you enough material to write a section, tell me what I'm missing rather than filling the gap with invented content.Why this works: The instruction to flag missing material (rather than fill gaps) is the most important constraint — it keeps you writing from your own evidence rather than AI-generated filler.
Good output: A draft analysis section interpreting your evidence against the literature, with hedged language, limitation acknowledgment, and explicit gaps flagged.
Compile an annotated bibliography
🟠 Claude
Here are all the sources I used in my paper: [LIST SOURCES WITH TITLES, AUTHORS, DATES, AND ANY NOTES] For each source: 1. Format the full citation in [APA 7th / MLA 9th / Chicago 17th / Harvard] style 2. Write a 3-sentence annotation: (a) what the source argues, (b) how I used it in my paper, (c) one limitation or bias I should be aware of 3. Flag: ✓ for sources I verified directly | ⚠ for sources I'm working from memory and should double-check At the end, list any claim in my paper that references a source I haven't included here — these are citation gaps I need to fill before submission. IMPORTANT: Only format the sources I've given you. Do not add sources I haven't listed, even if they seem relevant.
Why this works: The explicit 'only format what I give you' instruction prevents Claude from helpfully adding invented sources — which is the exact failure mode that has gotten students into academic integrity trouble.
Good output: A complete annotated bibliography in your chosen style, with verification flags, 3-sentence annotations, and a list of citation gaps in your paper.
Frequently Asked Questions
Will Claude make up sources for my research paper?
If you ask it to. These prompts are specifically designed to prevent that — they use CITATION PLACEHOLDER tags instead of asking Claude to generate references, and Step 5 explicitly instructs Claude to only format sources you provide. Follow the prompts as written.
Is using Claude to help write a research paper academic dishonesty?
That depends entirely on your institution's policy and how you use it. Using Claude to structure your argument, get feedback, and improve your draft is generally considered legitimate assistance. Submitting Claude-written text as your own work is dishonest. All citation placeholder tags in these prompts should be filled by you from your own research.
My paper needs very current research (2024–2026) — can Claude help?
Claude's training data has a cutoff date, so it may not know recent publications. For current research, use Gemini (which has web access) or a research database like Google Scholar directly. Claude is most useful for structuring your argument and drafting from evidence you've already gathered.
Make this workflow yours
Remix this pack into your workspace — edit the prompts, track your runs, and never hunt for them in old chats again.