Summarise a research paper (prompt)

curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/summarise-research-paper-prompt/raw' -o ~/'.claude/skills/summarise-research-paper-prompt/SKILL.md' && curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/summarise-research-paper-prompt/files/LICENSE.txt' -o ~/'.claude/skills/summarise-research-paper-prompt/LICENSE.txt'

curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/summarise-research-paper-prompt/raw' -o '.cursor/skills/summarise-research-paper-prompt/SKILL.md' && curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/summarise-research-paper-prompt/files/LICENSE.txt' -o '.cursor/skills/summarise-research-paper-prompt/LICENSE.txt'

curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/summarise-research-paper-prompt/raw' -o '.agents/skills/summarise-research-paper-prompt/SKILL.md' && curl -fsSL --create-dirs 'https://theskillvault.ai/api/skills/summarise-research-paper-prompt/files/LICENSE.txt' -o '.agents/skills/summarise-research-paper-prompt/LICENSE.txt'

{
  "mcpServers": {
    "theskillvault": {
      "url": "https://theskillvault.ai/mcp"
    }
  }
}

Papers bury the useful part. This prompt pulls out the question, the method, the results and the limits, in plain words, and says what it means for you.

When to use

• You have a paper (PDF text or an arXiv link) and limited time.
• You want to decide whether a paper is worth reading in full.
• You need to explain a paper to a team that does not read papers.

When not to use

• You want the key ideas from a talk, article or video. Use research-key-ideas-prompt.

Steps

1. Copy the paper text, or give an AI that can read PDFs the file or link.
2. Fill in {{PURPOSE}}: why you are reading it (for example "deciding whether to use this method in our search").
3. Paste it into the prompt below.
4. Check one or two of the numbers it reports against the paper.

Prompt

Summarise this paper for a smart reader who is not a specialist. My purpose: {{PURPOSE}}

1. Title, authors, year.
2. The question: what problem the paper tackles, in one or two sentences.
3. Core idea: the main idea or hypothesis in plain language.
4. Method: how they did it - the technique, model, data or experiment - without heavy maths.
5. What is new: what sets it apart from earlier work.
6. Results: the key findings, with the main numbers and what they were compared against.
7. Strengths and limits: what it does well, and its weaknesses, assumptions or gaps (small data, narrow tests, no code released).
8. So what for me: what this means for my purpose, and whether it is worth reading in full.

Only use what is in the paper. When you quote a number, say where in the paper it comes from. If a section is missing, say so.

PAPER:
{{PAPER TEXT}}

How to adapt

• Comparing papers? Run it on each, then ask for a table across sections 2, 4, 6 and 7.
• Need a tweet-length version? Add "finish with a two-sentence summary".

Gotchas

• Long papers can exceed an AI's input limit. Paste the abstract, method and results sections if so.
• Do not paste unpublished or confidential papers into a public AI tool without permission.

Portability

Works with any chat AI that accepts long text, in any field. No tool lock-in.

Done when

• Every section is filled or marked missing.
• The "so what" section answers your purpose, and the key numbers match the paper.

Example prompt

"Summarise this arXiv paper on retrieval-augmented generation. I am deciding whether to use its reranking method in our help bot."

---

Converted from the "summarize_paper" pattern in danielmiessler/Fabric (MIT, Copyright (c) 2012-2024 Scott Chacon and others). See LICENSE.txt.

Bundled files

• LICENSE.txt

Licence

Adapted from Daniel Miessler and Fabric contributors (summarize_paper pattern) (Copyright (c) 2012-2024 Scott Chacon and others), MIT License. Converted by TheSkillVault into a fill-in prompt: kept the goal, approach, novelty, results and limits structure, added a reader-purpose input, a so-what-for-me section and a quote-checking rule, steps and done-when sections.

Licence: MIT License. Attribution when you reuse it: Based on work by Daniel Miessler and Fabric contributors (summarize_paper pattern) (Copyright (c) 2012-2024 Scott Chacon and others, MIT License), https://github.com/danielmiessler/Fabric/blob/8ce0909d8622a61b6aaf539d467ba31bb7e0b6dd/data/patterns/summarize_paper/system.md; adapted by TheSkillVault, https://theskillvault.ai