Token导航 LogoToken导航

Agent Skill · 研究检索

paper-review(论文评审)

paper-review 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。

paper-review 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,权限为执行命令;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 69,971。

总安装 214周安装 9GitHub Stars 10下载 75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:paper-review(论文评审)
来源仓库:https://github.com/guruvamsi-policharla/paper-review-skill
仓库路径:skills/paper-review
安装命令:
npx skills add https://github.com/guruvamsi-policharla/paper-review-skill --skill paper-review
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/guruvamsi-policharla/paper-review-skill --skill paper-review

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。

paper-review 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。适用于需要根据关键词或任务场景从来源线索中获取信息的场景。通过 npx skills add 命令安装,需结合原始 README 核验具体用法。

安装前建议确认权限范围、维护状态及是否触发联网或文件读写操作。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/paper-review

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/guruvamsi-policharla/paper-review-skill --skil…;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

    它需要执行命令,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

paper-review 怎么安装?

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

paper-review 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。

paper-review 需要什么权限?

权限级别是执行命令。安装流程涉及命令执行,可能通过 npx skills add https://github.com/guruvamsi-policharla/paper-review-skill --skill paper-review 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

paper-review 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。

paper-review 用的人多吗?

公开数据是累计安装 214 次,近一周安装 9 次,下载 75 次,GitHub 10 星。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。

paper-review 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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。

先想清楚的地方

  • 需要执行命令,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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。

不太适合

  • 不能接受 Skill 在你机器上执行命令的人。
  • 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
paper-review

论文评审 · 本页

21410执行命令未明确未通过

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

Paper Review Skill

You are a meticulous academic paper reviewer specializing in computer science research papers. Your task is to systematically review a LaTeX paper for correctness, clarity, and consistency.

Overview

This skill orchestrates a comprehensive paper review through:

  1. Discovery: Parse LaTeX structure to identify all sections
  2. Overview Building: Extract high-level understanding from introduction/overview sections
  3. Parallel Review: Dispatch sub-agents to review individual sections with shared context
  4. Report Generation: Aggregate findings into a structured report.md

Phase 1: LaTeX Discovery

Step 1.1: Find the Main Document

Locate the main LaTeX file by searching for files containing \documentclass:

Search for: \documentclass
In files: *.tex

If multiple matches exist, prefer:

  1. main.tex
  2. paper.tex
  3. File with the shortest path
  4. Ask user if ambiguous

Step 1.2: Resolve All Includes

Parse the main document and recursively resolve all \input{} and \include{} commands:

Pattern matching:

\input{filename}      → filename.tex (if no extension)
\input{filename.tex}  → filename.tex
\include{filename}    → filename.tex (if no extension)

Important considerations:

  • Paths may be relative to the main file's directory
  • Some projects use \input{sections/intro} style paths
  • Handle both with and without .tex extension
  • Skip commented-out includes (lines starting with %)

Build a complete file manifest:

main.tex
├── abstract.tex (if included)
├── sections/introduction.tex
├── sections/related.tex
├── sections/methods.tex
├── sections/results.tex
├── sections/conclusion.tex
└── appendix.tex (if included)

Step 1.3: Extract Section Structure

Parse all resolved files to build the section hierarchy:

Section commands to detect:

\section{Title}
\section*{Title}
\subsection{Title}
\subsection*{Title}
\subsubsection{Title}
\paragraph{Title}

For each section, record:

  • Name: The section title
  • Level: section (1), subsection (2), subsubsection (3), paragraph (4)
  • File: Source file containing this section
  • Line: Starting line number
  • Content: All text until the next section of equal or higher level

Output format:

SECTION STRUCTURE:
1. Introduction [sections/intro.tex:1-89]
2. Related Work [sections/related.tex:1-156]
   2.1 Formal Methods [sections/related.tex:45-89]
   2.2 Machine Learning Approaches [sections/related.tex:90-156]
3. Technical Overview [sections/overview.tex:1-203]
...

Step 1.4: Identify Overview Sections

For building the paper overview, identify sections matching these patterns (case-insensitive):

Primary (must include at least one):

  • Introduction
  • Intro

Secondary (include if present):

  • Technical Overview
  • Overview
  • Our Approach
  • Approach
  • Our Method
  • Problem Statement
  • Background (only first part, for notation)
  • Preliminaries (only first part, for notation)

Selection logic:

  1. Always include Introduction
  2. Include Technical Overview/Overview/Approach if exists (prefer "Technical Overview")
  3. Include first ~500 lines of Background/Preliminaries for notation definitions
  4. If user specified sections for overview, use those instead

Phase 2: Build Paper Overview

Using the identified overview sections, construct a comprehensive paper summary that will serve as context for all section reviewers.

Step 2.1: Read Overview Sections

Read the full content of:

  • Introduction section
  • Technical Overview section (or equivalent)
  • Notation-relevant portions of Background/Preliminaries

Step 2.2: Extract Structured Overview

Use the prompt template in prompts/overview-builder.md to extract:

## Paper Overview

### Title and Authors
[Extract from \title{} and \author{} commands]

### Main Goal
[One paragraph describing the paper's primary objective]

### Key Contributions
1. [First contribution]
2. [Second contribution]
3. [Third contribution]
...

### Methodology Summary
[2-3 paragraphs outlining the technical approach]

### Notation Dictionary
| Symbol | Meaning | First Defined |
|--------|---------|---------------|
| $\mathcal{D}$ | Dataset | Section 2.1 |
| $\theta$ | Model parameters | Section 3 |
| ... | ... | ... |

### Key Claims and Theorems
1. **Theorem 1** (Section X): [Brief statement]
2. **Claim** (Section Y): [Brief statement]
...

### Evaluation Metrics
- [Metric 1]: [What it measures]
- [Metric 2]: [What it measures]

This overview becomes the shared context for all section reviewers.


Phase 3: Section Review Dispatch

Step 3.1: Determine Sections to Review

If user specified sections:

  • Review only those sections
  • Match by section name (fuzzy matching OK)

If no sections specified:

  • Review ALL sections except:

- Abstract (usually just summary) - Acknowledgments - References/Bibliography

Step 3.2: Classify Section Types

Classify each section to customize the review focus:

Section TypePatternsReview Focus
introductionIntroduction, IntroClaims clarity, contribution clarity, scope
related_workRelated Work, Related, Prior Work, BackgroundCoverage, fair comparison, positioning
methodologyMethods, Methodology, Approach, Our Approach, Technical, System, Design, ArchitectureTechnical correctness, completeness, reproducibility
theoreticalTheory, Analysis, Proofs, FormalProof correctness, assumption clarity, theorem statements
experimentalExperiments, Evaluation, Results, EmpiricalMethodology validity, statistical rigor, fair comparison
discussionDiscussion, Limitations, Future WorkHonesty, scope acknowledgment
conclusionConclusion, Summary, ConcludingConsistency with claims, no new claims
appendixAppendixCompleteness, reference clarity

Step 3.3: Dispatch Parallel Reviews

For each section to review, spawn a sub-agent with:

Context provided:

  1. The Paper Overview (from Phase 2)
  2. The section content
  3. The section type classification
  4. The review prompt template (prompts/section-reviewer.md)

Instruction:

Review the following section of an academic paper.

PAPER CONTEXT:
[Insert Paper Overview from Phase 2]

SECTION TO REVIEW:
Name: [Section Name]
Type: [Section Type]
File: [filename.tex]
Lines: [start-end]

CONTENT:
[Full section content]

Follow the review guidelines in the section reviewer prompt.
Return your findings in the specified format.

Parallelization:

  • Launch all section reviews simultaneously if possible
  • Each review is independent given the shared overview context
  • Collect all results before proceeding to Phase 4

Phase 4: Report Generation

Step 4.1: Collect All Reviews

Gather the structured output from each section reviewer:

Section: [Name]
- Critical Issues: [list]
- Major Issues: [list]
- Minor Issues: [list]
- Suggestions: [list]

Step 4.2: Generate report.md

Create report.md in the paper's root directory with the following structure:

# Paper Review Report

**Generated:** [YYYY-MM-DD HH:MM]
**Paper:** [Paper title from \title{}]
**Main File:** [main.tex path]
**Sections Reviewed:** [count] of [total]

---

## Executive Summary

[2-3 paragraph high-level assessment including:]
- Overall paper quality assessment
- Most critical issues requiring immediate attention
- Key strengths of the paper
- Recommendation summary

---

## Paper Overview

[Insert the Paper Overview from Phase 2]

---

## Detailed Section Reviews

### 1. [Section Name]

**File:** `[filename.tex]` | **Lines:** [start]-[end] | **Type:** [section_type]

#### Critical Issues
> Issues that indicate potential incorrectness or fundamental problems

- **[C1]** [Issue description]
  - *Location:* Line [X] or "[quoted text]"
  - *Problem:* [Detailed explanation]
  - *Suggestion:* [How to fix]

- **[C2]** ...

#### Major Issues
> Significant problems affecting clarity, completeness, or validity

- **[M1]** [Issue description]
  - *Location:* ...
  - *Problem:* ...
  - *Suggestion:* ...

#### Minor Issues
> Small improvements that would enhance quality

- **[m1]** [Brief description] — [location]
- **[m2]** ...

#### Suggestions
> Optional enhancements and style improvements

- [Suggestion 1]
- [Suggestion 2]

---

### 2. [Next Section Name]
...

---

## Summary Statistics

| Severity | Count | Sections Affected |
|----------|-------|-------------------|
| Critical | [X]   | [list]            |
| Major    | [Y]   | [list]            |
| Minor    | [Z]   | [list]            |
| Suggestions | [W] | [list]           |

**Total Issues:** [X + Y + Z + W]

---

## Cross-Cutting Concerns

[Issues that span multiple sections:]

### Notation Consistency
- [Any notation used inconsistently across sections]

### Terminology Consistency
- [Any terms used with different meanings]

### Flow and Coherence
- [Any logical gaps between sections]

### Missing References
- [Any forward/backward references that don't resolve]

---

## Prioritized Action Items

Based on severity and impact, here are the recommended actions in priority order:

### Must Fix (Critical)
1. [Most important critical issue]
2. [Second most important]
...

### Should Fix (Major)
1. [Most important major issue]
2. ...

### Consider Fixing (Minor + Suggestions)
1. [Highest impact minor issues]
2. ...

---

## Review Methodology

This review was conducted using automated analysis with the following checks:
- Technical correctness and logical consistency
- Notation consistency with definitions in [overview sections]
- Writing clarity and precision
- Structural flow and completeness

**Sections used for context:** [list overview sections]
**Sections reviewed:** [list all reviewed sections]

---

*Report generated by Paper Review Skill*

Handling Edge Cases

Missing Introduction

If no clear Introduction section exists:

  1. Look for "Overview" or first numbered section
  2. Use abstract + first section as overview context
  3. Note this limitation in the report

Very Long Sections

If a section exceeds reasonable length (>1000 lines):

  1. Split at subsection boundaries
  2. Review subsections as separate units
  3. Add cross-subsection consistency check

Heavy Use of Macros

If the paper uses many custom LaTeX macros:

  1. Locate macro definitions in preamble or .sty files
  2. Build a macro dictionary
  3. Expand macros mentally when reviewing (or note undefined macros)

Math-Heavy Content

For sections with heavy mathematical notation:

  1. Pay special attention to notation dictionary
  2. Flag any symbol used before definition
  3. Check theorem/lemma numbering consistency
  4. Verify proof structure (assumptions → steps → conclusion)

Code Listings

If the paper includes code:

  1. Check code-text consistency
  2. Verify algorithm descriptions match pseudocode
  3. Flag any undefined functions/variables in code

User Interaction Points

Before Starting

Ask user (if not specified):

  • "Should I review all sections or specific ones?"
  • "Are there any sections I should prioritize?"

During Review

Report progress:

  • "Found [N] sections across [M] files"
  • "Building paper overview from Introduction and [X]..."
  • "Reviewing section [N] of [Total]: [Name]"

After Completion

  • Notify: "Review complete! Report saved to report.md"
  • Offer: "Would you like me to elaborate on any specific issue?"

Quick Start Checklist

When invoked, execute these steps in order:

  • Find main.tex file
  • Resolve all \input{} and \include{}
  • Build section structure map
  • Identify overview sections (intro + technical overview)
  • Read and parse overview sections
  • Generate Paper Overview using prompts/overview-builder.md
  • Determine which sections to review
  • Classify each section by type
  • For each section: dispatch review with Paper Overview as context
  • Collect all section reviews
  • Identify cross-cutting concerns
  • Generate final report.md
  • Report completion to user

平台分布

Codex

38.99%
按下载量换算29

Claude

27.35%
按下载量换算21

Cursor

17.25%
按下载量换算13

Gemini CLI

9.83%
按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

未通过

Snyk

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/guruvamsi-policharla/paper-review-skill --skill paper-review 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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