Paper Reader

Comprehensive PDF paper reader for academic research. Extracts text, figures, tables, and structured content from research papers with support for multimodal...

Security Scan

The skill's description promises a CLI/Python library for extracting text, figures, and tables, but the package provides no code or install instructions — the runtime instructions reference scripts and modules that are not included, which is an incoherence you should resolve before installing or running anything.

Purpose & Capability

The SKILL.md describes a full CLI and Python API and lists Python dependencies (pdfplumber, pymupdf, Pillow) that are appropriate for a PDF reader, but the package contains no code files (no read_paper.py, no paper_reader module). That mismatch means the skill can't perform the claimed capabilities as distributed.

Instruction Scope

Runtime instructions explicitly tell the agent/user to run python3 ~/.openclaw/skills/paper-reader/read_paper.py and to import paper_reader — paths and modules that do not exist in this bundle. The instructions also direct reading PDFs and writing extracted outputs to arbitrary filesystem locations (e.g., ~/..., ./figures), which is expected for a paper-processing tool but should only be done by code you can inspect. There is no guidance about network activity (no explicit exfiltration), but the missing code makes it impossible to verify behavior.

Install Mechanism

There is no install spec (instruction-only), which is lower risk in itself. The README suggests running pip install pdfplumber pymupdf Pillow manually; that is reasonable for the declared dependencies, but because no code is bundled, there's nothing to install or verify from this skill package.

Credentials

The skill does not request environment variables, credentials, or config paths. The declared Python packages are appropriate for the stated purpose and do not imply additional secrets or cloud access.

Persistence & Privilege

The skill does not request always:true or other elevated persistent privileges. Model invocation is enabled by default (normal). The skill does not attempt to modify other skills or system-wide settings in the included materials.

What to consider before installing

Do not run the example commands or pip-install things on the basis of this package alone. The SKILL.md references scripts and a Python module (read_paper.py, paper_reader) that are not included — ask the publisher for the missing code or a proper install spec before installing. If you still want to use it, request: (1) the actual code files or a reproducible install (GitHub repo or PyPI package), (2) a clear statement of whether the tool performs any network calls and to which endpoints, and (3) a signed or verifiable author/source. If you must test now, do so in a restricted sandbox and inspect any downloaded/extracted files before running them. Because the package is incoherent as distributed, treat it as untrusted until the missing pieces are resolved.

Paper Reader Skill

专为学术论文阅读设计的多功能PDF提取工具。

功能特性

• 📄 文本提取: 完整提取论文文本，保留章节结构
• 🖼️ 图片提取: 自动提取论文中的图表、实验结果图
• 📊 表格提取: 提取实验数据表格为结构化格式
• 📋 元数据读取: 标题、作者、摘要、关键词
• 🔍 章节识别: 自动识别Introduction、Methods、Results等章节
• 📑 批量处理: 支持整个论文文件夹批量处理

依赖安装

pip install pdfplumber pymupdf Pillow

使用方法

命令行工具

# 完整提取论文（文本+图片+表格）
python3 ~/.openclaw/skills/paper-reader/read_paper.py paper.pdf --full

# 仅提取文本
python3 ~/.openclaw/skills/paper-reader/read_paper.py paper.pdf --text

# 提取图片到指定文件夹
python3 ~/.openclaw/skills/paper-reader/read_paper.py paper.pdf --images --img-dir ./figures

# 提取表格为CSV
python3 ~/.openclaw/skills/paper-reader/read_paper.py paper.pdf --tables --csv-dir ./tables

# 提取前3页（通常包含摘要和引言）
python3 ~/.openclaw/skills/paper-reader/read_paper.py paper.pdf --pages 1-3

Python API

from paper_reader import PaperReader

reader = PaperReader("paper.pdf")

# 获取元数据
metadata = reader.get_metadata()

# 提取全文
text = reader.extract_text()

# 提取特定章节
abstract = reader.extract_section("Abstract")
methods = reader.extract_section("Methods")

# 提取图片
figures = reader.extract_images(output_dir="./figures")

# 提取表格
tables = reader.extract_tables()

输出结构

paper_output/
├── paper.txt              # 完整文本
├── metadata.json          # 论文元数据
├── abstract.txt           # 摘要
├── sections/              # 分章节文本
│   ├── 01_Introduction.txt
│   ├── 02_Related_Work.txt
│   ├── 03_Methods.txt
│   ├── 04_Experiments.txt
│   └── 05_Conclusion.txt
├── figures/               # 提取的图片
│   ├── fig_1.png
│   ├── fig_2_table.png
│   └── fig_3_chart.png
└── tables/                # 提取的表格
    ├── table_1.csv
    └── table_2.csv

学术论文专用功能

1. 实验数据提取

自动识别并提取：

• 实验结果表格
• 性能对比图表
• 消融实验数据

2. 方法章节结构化

识别Methods章节中的：

• 网络架构描述
• 损失函数公式
• 训练参数设置

3. 参考文献提取

提取参考文献列表，支持BibTeX格式导出。