Academic Prompt Library
TL;DR
- 📚 Resource directory: 20+ curated open-source/official academic prompt libraries, grouped by scenario (reading / review / writing / peer review / submission), all in tables with links and notes
- 🤖 Ready to use: pair with PaperPop's Send to AI — add any prompt below to a custom prompt folder, then send the current paper's title/abstract/URL to ChatGPT, Claude, DeepSeek, or Kimi in one click
- 🧩 Concepts clarified: a full worked example of how a Skill relates to a Prompt, and how to convert between them
1. How to use this with PaperPop
PaperPop's "Send to AI" automatically detects the current paper's Title, Abstract, and URL and appends them to the prompt you choose, then copies everything and jumps to the AI platform. So when adding prompts from the libraries below, keep only the instruction part — you don't need to include the paper content, because PaperPop fills that in for you.
Recommended workflow:
- In PaperPop settings, create folders (e.g. "Reading", "Writing", "Review") matching the scenarios below;
- Pick a prompt from the resources here, copy its instruction text, and save it as a custom prompt;
- On a paper's detail page, tick the fields you want (Title / Abstract / URL), pick the target AI, and send in one click.
Note: all custom prompts are stored locally in your browser. Back up the ones you rely on.
2. General prompt libraries (copy-paste, ready to use)
These repos are written for "paste straight into the chat box", which makes them ideal for importing into PaperPop.
| Name | Language | Link | Notes |
|---|---|---|---|
| ahmetbersoz / chatgpt-prompts-for-academic-writing | EN | GitHub | One of the most-starred academic-writing prompt collections; covers ideation, lit review, paraphrasing, and research planning; regularly updated |
| xuhangc / ChatGPT-Academic-Prompt | EN | GitHub | Role-play style academic prompts, e.g. "write a paper review as a senior scientist" |
| xstongxue / best-prompts | ZH | GitHub | Scenario-curated Chinese paper-writing prompts for the chat box, with multiple ranked candidates per task |
3. Paper reading & interpretation
Good for a "Reading" folder in PaperPop: summarize, critically read, and extract key points from the current paper.
| Name | Language | Link | Notes |
|---|---|---|---|
| Anthropic Prompt Library | EN | docs.anthropic.com | Official curated prompts; "Cite your sources" is useful for interpreting papers with grounded citations |
| FIU Library · Literature Reviews with Prompts | EN | library.fiu.edu | University-library guide demonstrating a full step-by-step workflow for speeding up reading and review |
| AI Academy · 35 Research Prompts | EN | academy.techpresso.co | Grouped by literature analysis / research design / data analysis, with explicit caveats (don't rely on it for citations) |
4. Academic writing, polishing & translation
| Name | Language | Link | Notes |
|---|---|---|---|
| xstongxue / best-prompts (paper writing) | ZH | GitHub | Structural-imitation and abstract-writing prompts that rewrite drafts along a model paper's logical skeleton |
| writing-resources / awesome-scientific-writing | EN | GitHub | A curated list of tools, templates, and workflows for scientific writing beyond LaTeX — a companion to prompts |
| ClickUp · AI Prompts for Literature Review | EN | clickup.com | Practical literature-review prompts for ChatGPT/Gemini/Perplexity, framed around review tasks |
5. Peer review & submission
| Name | Language | Link | Notes |
|---|---|---|---|
| xuhangc / ChatGPT-Academic-Prompt | EN | GitHub | Includes a "senior-scientist paper review" template, ideal for mock peer review |
| Research Prompt Pro | ZH | Web tool | Form-driven generator for cover letters, reviewer-response letters, email replies, and reviewer personas |
| bohyy / academic-ai-prompt | ZH | GitHub | A full grad-student prompt library: topic selection, search, review, and top-journal writing (incl. rebuttals) |
6. Chinese-focused & searchable collections
| Name | Language | Link | Notes |
|---|---|---|---|
| bohyy / academic-ai-prompt | ZH | GitHub | 89+ prompts for master's/PhD students across the full pipeline, with an evaluation spreadsheet and templates |
| wenhaomin / ChatGPT-PromptGenius | EN/ZH | GitHub | A general prompt site with hierarchical categories, keyword search, one-click copy, and multi-language support |
| yunwei37 / Prompt-Engineering-Guide-zh-CN | ZH | GitHub | A continuously updated Chinese prompt-engineering resource hub (guides, papers, lectures, notebooks) |
7. Official resources & prompt-engineering guides
| Name | Language | Link | Notes |
|---|---|---|---|
| Anthropic Prompt Library | EN | docs.anthropic.com | Claude's official library: 60+ prompts grouped by business/personal tasks, demonstrating XML-structured formatting |
| Anthropic Prompt Engineering Overview | EN | docs.claude.com | Official guide: clear instructions, positive/negative examples, step-by-step reasoning, XML tags |
| Claude Code Prompt Library | EN | code.claude.com | Copy-paste prompts tagged by task/role, each with a "why this works" pattern explanation |
8. Advanced: Agent Skill libraries
The resources below are Skills, not plain prompts — they live as SKILL.md files invoked automatically by agent tools (Claude Code, Cursor, Codex, OpenClaw). This differs from PaperPop's copy-paste flow, but the contents of a SKILL.md can be "flattened" into a paste-ready prompt (see the next section).
| Name | Language | Link | Notes |
|---|---|---|---|
| xstongxue / best-skills | ZH | GitHub | A general Skills collection; the README clearly contrasts "chat-box Prompts" with "agent-facing SKILL.md" |
| alfonso0512 / research-writing-skill | EN/ZH | GitHub | 30 paper-writing prompt templates, also packaged as a Skill; README explains how to copy templates into any AI tool |
| Norman-bury / research-writing-skill | ZH | GitHub | Modular research-writing skill: review, chapter writing, figures, peer review, statistical analysis, and more |
| zLanqing / codex-claude-academic-skills | ZH | GitHub | Three Skills for Chinese researchers: reading reports/PPTs, writing/polishing/review, and scientific computing/figures |
| LeonChaoX / qinyan-academic-skills | EN/ZH | GitHub | 183 academic agent Skills across 18 areas: search, writing, review, grant proposals (NSFC/NSSFC), and more |
| Yuan1z0825 / nature-skills | EN/ZH | GitHub | Skills for Nature-style academic phrasing and scientific figure-making |
| HughYau / AcademicForge | EN/ZH | GitHub | A one-stop Skills platform for academic writing; pick skill packs on the web and export per platform |
9. PaperPop Custom Prompt Guide
Method 1 · Auto-append (default) Write only the instruction. The paper's "Title / Abstract / Link" will be automatically appended to the end of the instruction based on the checkboxes selected in the send panel — no manual entry needed.
Method 2 · Template Variables (precise position control) Insert {{title}}, {{abstract}}, {{url}} into your instruction, and they will be replaced with the corresponding content when sending. Once template variables are used, the "Include Content" checkboxes in the send panel will be ignored, and the prompt takes full precedence.
⚠️ Usage caveat: AI may fabricate citations and data. Always verify specific references and statistics against the original text and databases.
10. Skills vs. Prompts: the relationship and how to convert
10.1 What each one is
- Prompt: a one-off instruction. You paste it into the chat box, the model runs it on the spot, and it's gone when you close the window — next time you have to find and paste it again. PaperPop's "Send to AI" handles Prompts.
- Skill: a structured, reusable capability package with a trigger condition, centered on a
SKILL.mdfile (YAML frontmatter + body), often with reference docs, scripts, and templates. It lives in an agent'sskills/directory (Claude Code, Cursor, Codex), and the agent decides automatically when to invoke it based on keywords in thedescription— no manual pasting each time.
In one line:
Skill ≈ Prompt template + trigger condition + context/tool configuration + optional scripts and resource files. A Prompt is "how to do it this time"; a Skill is "do it this way every time this kind of task comes up."
The xstongxue/best-skills README puts it well: its companion prompts are chat-box Prompts you copy-paste manually, while its Skills are agent-facing SKILL.md files the agent invokes automatically based on keywords and trigger scenarios.
10.2 What a SKILL.md looks like
Here is a minimal "critical paper reading" Skill:
---
name: critical-paper-review
description: Use when the user provides a paper's title, abstract, or
full text and wants a critical read pointing out methodological
limitations and potential bias. For peer review, lab meetings,
and close reading.
---
# Critical Paper Reading
## Role
You are a senior peer reviewer in this field. Stay objective and
distinguish the authors' claims from established conclusions.
## Steps
1. Summarize the research question, method, and main conclusion
(one sentence each).
2. Point out potential methodological limitations or confounders.
3. Assess whether the conclusions are well supported and whether
there is overclaiming.
4. List three key questions you would ask the authors.
## Output format
Four headed sections: Overview / Limitations / Evidence / Questions.
## Constraints
- Do not fabricate data or citations; flag any uncertainty.It has two parts: the YAML frontmatter (name names the skill; description tells the agent when to use it, which is the key to auto-triggering) and the body (role, steps, output format, constraints — the actual instructions).
10.3 Skill → Prompt: flattening a Skill into a paste-ready prompt
PaperPop uses the "paste into the chat box" route and won't auto-load a SKILL.md. To use a Skill inside PaperPop, just flatten it into one prompt:
- Drop the YAML frontmatter —
name/descriptionexist for agent auto-routing and aren't needed when pasting manually. - Keep the body — role, steps, output format, and constraints are the prompt itself.
- Remove references to external files/scripts/tools — there are none in a chat box, so inline anything essential.
- Remove paper-content placeholders — PaperPop appends the title/abstract/URL, so don't write "paste the paper here."
Flatten the Skill above and you get exactly the ② Critical reading prompt in Section 9 — just save it as a PaperPop custom prompt.
10.4 Prompt → Skill: promoting a useful prompt into a Skill
Conversely, if you reuse a PaperPop prompt constantly and also work in an agent like Claude Code or Cursor, you can promote it to a Skill to gain auto-triggering, team sharing, and version control:
- Add YAML frontmatter — give it a
nameand write a gooddescriptioncontaining trigger keywords and use cases (this determines whether the agent can find it automatically). - Separate process from background — keep the "how to do it" steps in
SKILL.md, and move lengthy background, examples, and templates into reference files (e.g.references/) so the main file stays lean and key instructions aren't buried. - Add constraints and an output contract — define what "done" means, the output format, and any required sections, so it doesn't become a "fair-weather skill" that works in the demo but breaks on the third real use.
- Drop it in the agent's skills directory — then natural language ("critically read this paper for me") triggers it automatically.
10.5 Which should you use?
| Scenario | Better fit |
|---|---|
| Reading papers in the browser, asking AI ad hoc | Prompt (with PaperPop's one-click send) |
| The same kind of task repeated often, want consistency | Skill |
| Team collaboration needing versioning and sharing | Skill |
| Working only in a web chat box, no coding agent | Prompt |
| Tasks needing scripts, file I/O, multi-step branching | Skill (or even an Agent) |
In short: inside PaperPop, what you need is a Prompt. When a prompt becomes a fixed move in your workflow and you happen to work in an agent environment, that's when to consider promoting it to a Skill.
Related links
- Send to AI: add these prompts to a custom folder and send in one click
- Abstract Interpretation: get multi-angle interpretation in the browser
- Mind Map: turn a paper into a mind map
Note: All listed resources come from the public web; links and contents may change over time — refer to each repo/site's latest version. The example prompts in Section 9 are original and free to use and modify. When using AI to interpret papers, follow your institution's academic-integrity and AI-use policies, and verify citations, data, and other critical details yourself.