ljg-paper
The ljg-paper skill processes research papers, specifically in the field of large models and AI, to extract actionable insights for non-academic readers. It translates technical content into a structured 'world-model update' consisting of six sections: quick summary, critique of existing beliefs, new claims, credibility assessment, practical application rules, and limitations. It generates org-mode notes saved to the local file system.
Is ljg-paper safe to install?
Review before installing: our audit of ljg-paper's source files found 2 shell commands, 0 external URLs, file reads and writes (medium risk). Every command and URL listed appears verbatim in the skill's source. The skill reads local paper files and writes generated org-mode notes to the ~/Documents/notes/ directory. It executes shell commands to generate timestamps for file naming.
How we audit skills: our security review methodology.
Who is this skill for?
Non-academic readers, researchers, and practitioners who want to understand the implications of AI research papers without needing to reproduce experiments or perform formal peer reviews.
What can you do with it?
- Summarizing and analyzing research papers from arxiv links, PDFs, or local files.
- Explaining the core arguments of a paper based on its title.
- Generating structured org-mode notes for future reference.
- Providing oral explanations of research papers for users.
How good is this skill?
Quality score: 9/10. The skill provides a highly structured and clear workflow for distilling complex research into actionable insights. The instructions are specific, and the constraints on output format and content are well-defined. The risk level is accurately identified as medium due to file system interaction.
What does the skill file contain?
# ljg-paper: 把论文翻译成世界模型更新 这不是论文精读器，也不是审稿器。它只做一件事：把一篇大模型相关论文，翻译成一个非专业人士能带走的判断。 读者不需要知道完整实验、数据细节、公式推导、模型结构图。读者要知道：这篇论文让我对大模型的哪条看法改了一点，改得有多稳，不能拿去误用到哪里。 ## Workflow Routing | 输入 | 做法 | |------|------| | arxiv / PDF / paper URL / 本地论文 | 读论文，只抓能回答六个问题的信息 | | 只有论文标题 | 先找到论文，再按同一框架读 | | 用户说「讲论文」「读论文」「paper」「分析这篇」 | 默认生成 org 笔记并保存 | | 用户只要口头解释 | 不写文件，按同一六节口头讲 | ## Gotchas - 不要把「非专业人士」误解成「加很多类比」。类比只在能改变理解时用；装饰性类比会拖慢读者。 - 不要复述论文结构。Abstract / Introduction / Method / Experiments 的顺序不是读者的理解顺序。 - 不要强制讲方法。方法只在它解释了「新想法为什么可能成立」时出现。 - 不要强制讲数字。数字只在它会改变可信度判断时出现；最多挑 1-2 个关键数字。 - 不要做博导审稿。这里判断「我该信几分」，不判断「该不该接收」...
Frequently asked questions
Does this skill perform formal peer reviews or reproduce experiments?
No. The skill explicitly avoids formal peer review, exhaustive method summaries, benchmark tables, and experiment reproduction.
How does the skill handle technical jargon and formulas?
It avoids academic jargon and LaTeX formulas in the main text, translating them into natural language to ensure accessibility for non-experts.
Where are the generated notes saved?
The skill saves notes in the ~/Documents/notes/ directory using a specific timestamped filename format.
What is the primary goal of the output?
The goal is to provide a 'world-model update' that helps the reader change their judgment about large models, identifying what to believe and where the paper's claims might break.
Related skills
azure-ai
429.4KDevelopers and AI agents requiring integration with Azure AI services for search, transcription, and synthesis tasks
The azure-ai skill provides tools for interacting with Azure AI services, including AI Search, Speech, OpenAI, and Document Intelligence. It supports search operations, speech-to-text, text-to-speech, and OCR through an MCP server interface.
image-to-video
355.0KDevelopers and users who need to automate video generation from images using the RunComfy platform
The image-to-video skill routes user requests to specific RunComfy animation models based on intent. It selects HappyHorse 1.0 I2V for general animation, Wan 2.7 for custom-voiceover lip-sync, or Seedance 2.0 Pro for multi-modal composition. The skill executes the RunComfy CLI to process inputs and download generated video files.
face-swap
245.4KUsers needing to perform face or character swaps on media assets via the RunComfy CLI
The face-swap skill provides a CLI-based interface to perform identity replacement on images and videos using various RunComfy model endpoints. It routes requests to specific models based on user intent, such as audio-driven animation, motion transfer, batch processing, or high-fidelity still image editing.
paper-context-resolver
140.4KDevelopers and researchers performing deep learning repository reproduction who encounter specific technical ambiguities in documentation
The paper-context-resolver skill provides targeted information from research papers to resolve specific gaps in deep learning repository reproduction. It focuses on technical details like dataset splits, preprocessing steps, evaluation protocols, checkpoint mappings, and runtime assumptions.