Token导航 LogoToken导航

Agent Skill · 研究检索

scientific-paper(科学论文)

scientific-paper 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

scientific-paper 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 46,384。

总安装 783周安装 32GitHub Stars 公开资料未说明下载 251核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:scientific-paper(科学论文)
来源仓库:https://github.com/jcardif/agent-skills
仓库路径:skills/scientific-paper
安装命令:
npx skills add https://github.com/jcardif/agent-skills --skill scientific-paper
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/jcardif/agent-skills --skill scientific-paper

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

scientific-paper 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。适用于科研选题、文献调研、信息聚合等需要从多源数据中提炼线索的场景。

通过安装命令 npx skills add https://github.com/jcardif/agent-skills --skill scientific-paper 添加,需确认权限范围和维护状态。使用前应核实是否会触发联网、命令执行或文件读写操作,避免越权访问。

建议结合原始 README 进一步核验具体用法和功能边界。

来源仓库内路径:skills/scientific-paper

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/jcardif/agent-skills --skill scientific-paper;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

scientific-paper 怎么安装?

在终端执行 npx skills add https://github.com/jcardif/agent-skills --skill scientific-paper。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

scientific-paper 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

scientific-paper 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

scientific-paper 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

scientific-paper 用的人多吗?

公开数据是累计安装 783 次,近一周安装 32 次,下载 251 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

scientific-paper 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
scientific-paper

科学论文 · 本页

783—只读未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Scientific Paper Writing

Core philosophy

Approach every paper as a PhD-level researcher would:

  1. Understand before writing. Read existing content, the problem domain, and related work before producing a single sentence.
  2. Rigor over volume. Every claim must be substantiated — by derivation, citation, or empirical evidence. Never pad.
  3. Precision in language. Use exact terminology. Define terms on first use. Avoid weasel words ("very", "clearly", "obviously").
  4. Reproducibility. A competent reader should be able to reproduce every result from what is written.
  5. Intellectual honesty. State assumptions explicitly. Acknowledge limitations. Report negative results. Distinguish what is proven from what is conjectured.
  6. Write to communicate, not to impress. The goal is transferring understanding to the reader. Prefer simple, direct prose over elaborate vocabulary. Complex ideas demand simple writing.

Research methodology workflow

When asked to write or substantially expand a paper, follow this process:

Phase 1: Problem Formulation
  - What is the precise research question?
  - Why does it matter? (motivation)
  - What would a solution look like? (success criteria)
  - What are the boundaries? (scope)

Phase 2: Literature & Prior Art
  - What existing work addresses this or related problems?
  - What are the gaps in existing approaches?
  - How does this work position itself relative to the field?

Phase 3: Methodology & Framework
  - What analytical/computational approach will be used?
  - Why is this approach appropriate? (justify the choice)
  - Define notation, assumptions, and formal setup

Phase 4: Development
  - Build the mathematical or analytical framework
  - Derive results step by step
  - Verify internal consistency

Phase 5: Analysis & Discussion
  - What do the results mean?
  - What are the limitations?
  - What are the practical implications?

Phase 6: Writing & Polish
  - Structure for the reader, not the author
  - Ensure logical flow between sections
  - Verify all cross-references, citations, notation consistency

For detailed guidance on each phase, see references/scientific-method.md.

LaTeX standards

Document setup

Always use these baseline packages unless there is a specific reason not to:

\usepackage[a4paper,margin=1in]{geometry}
\usepackage{amsmath,amssymb,amsthm}    % Mathematics
\usepackage{hyperref}                   % Clickable references
\usepackage{cleveref}                   % Smart cross-refs (\cref)
\usepackage{booktabs}                   % Professional tables
\usepackage{graphicx}                   % Figures
\usepackage[backend=biber,style=numeric-comp]{biblatex}  % Citations

Mathematical typesetting rules

  • Operators: Use \operatorname{} or define with \DeclareMathOperator for multi-letter operators (e.g., \operatorname{argmax}, not argmax in math mode)
  • Variables: Single italic letters for scalars ($x$), bold for vectors ($\mathbf{v}$), blackboard bold for sets ($\mathbb{R}$)
  • Subscripts/superscripts: Use descriptive subscripts: $r_{\text{net}}$, not $r_n$ when $n$ is ambiguous
  • Equations: Number all important equations. Use align for multi-line, equation for single-line. Never use eqnarray.
  • Definitions: Use \coloneqq (:=) when defining a quantity, = for equality
  • Notation table: Include a notation summary table when the paper uses more than 10 symbols

Theorem environments

Define and use proper theorem environments:

\newtheorem{theorem}{Theorem}[section]
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{corollary}[theorem]{Corollary}
\theoremstyle{definition}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{example}[theorem]{Example}
\theoremstyle{remark}
\newtheorem{remark}[theorem]{Remark}

Tables

Use booktabs rules (\toprule, \midrule, \bottomrule). Never use vertical lines. Example:

\begin{table}[htbp]
  \centering
  \caption{Description before the table.}
  \label{tab:example}
  \begin{tabular}{lrr}
    \toprule
    Item & Value & Unit \\
    \midrule
    Precision & 94.3 & \% \\
    Samples   & 1{,}000 & -- \\
    \bottomrule
  \end{tabular}
\end{table}

For complete LaTeX reference, see references/latex-reference.md.

Writing quality standards

  • Active voice preferred over passive ("We derive..." not "It is derived...")
  • Present tense for established facts and mathematical truths; past tense for describing what was done
  • One idea per paragraph. Topic sentence first, then support.
  • Transitions between sections and paragraphs — the reader should never wonder "why am I reading this now?"
  • Avoid: "it is well known that", "trivially", "clearly", "the reader can easily verify". If it is clear, the reader does not need to be told; if it is not, saying so is condescending.
  • Define before use. Every symbol, abbreviation, and technical term must be defined before or at first use.

For detailed style guidelines, see references/writing-style.md.

Workflow for editing existing papers

When asked to work on an existing .tex file:

  1. Read the entire file first. Understand structure, notation, and what exists.
  2. Identify what is needed. Expand content? Fix errors? Add sections? Improve prose?
  3. Maintain consistency. Match existing notation, style, and conventions. Do not introduce new packages or notation that conflicts with what exists.
  4. Build on what is there. Extend, do not replace, unless explicitly asked to rewrite.
  5. Verify compilation. Ensure no LaTeX errors are introduced.

Workflow for new papers

  1. Clarify the research question with the user before writing.
  2. Outline first. Propose a section structure and get agreement before writing prose.
  3. Design figures and tables early. Determine the key visuals before writing prose — let the figure sequence drive the narrative structure.
  4. Write non-linearly. Start with Results/Methods (the most concrete sections), then Discussion, then Introduction, then Abstract. Few successful papers are written start-to-finish.
  5. Revise in order: content → structure → style. First ensure correctness and completeness, then logical organization, then prose quality.
  6. Writing is iterative. Drafting reveals gaps in reasoning. Expect to revisit earlier sections as later ones develop.

Quality checklist

Before considering a paper section complete:

- [ ] Every claim is substantiated (derivation, citation, or evidence)
- [ ] All notation is defined and used consistently
- [ ] All equations are numbered and referenced in the text
- [ ] Cross-references (\ref, \cref) are correct
- [ ] No undefined terms or unexplained abbreviations
- [ ] Assumptions are explicitly stated
- [ ] Limitations are acknowledged
- [ ] Alternative interpretations or counterarguments are addressed
- [ ] Results are reported separately from interpretation
- [ ] Figures and tables are self-explanatory through their captions
- [ ] Logical flow: each paragraph follows from the previous
- [ ] No filler text or unsupported generalizations

平台分布

Codex

36.47%
按下载量换算92

Claude

29.8%
按下载量换算75

Cursor

19.15%
按下载量换算48

Gemini CLI

9.89%
按下载量换算25

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills