一个 Claude Code skill，将技术文章转换为易懂内容 — 把充满术语的 AI/科技新闻变成任何人都能理解的大白话。
Easy Read（人话翻译器） 帮助非技术人员理解复杂的技术内容。它自动识别难懂的术语，创建美观的词汇表，用大白话解释，并提供文章摘要，让完全没有技术背景的人也能看懂。
核心特点：
https://github.com/user-attachments/assets/[demo-video-placeholder]
- 零技术背景要求 — 为任何对科技新闻好奇的人设计
- 智能术语识别 — 自动识别需要解释的术语（AI 概念、缩写、流行词）
- 美观的输出 — 简洁现代的 HTML 词汇表，带难度等级、发音指南和历史背景
- 多种输入格式 — 接受文本、文件、网址、PDF、截图 — 任何你扔给它的东西
- 信息平权 — 每个人都应该理解正在塑造我们世界的技术，无论背景如何
直接从 Claude Code 安装：
/plugin marketplace add HeiGeAi/easy-read
/plugin install easy-read@easy-read然后在 Claude Code 中输入 /easy-read 使用。
将 skill 文件复制到你的 Claude Code skills 目录：
# 克隆仓库
git clone https://github.com/HeiGeAi/easy-read.git ~/.claude/skills/easy-read或手动复制：
# 创建 skill 目录
mkdir -p ~/.claude/skills/easy-read/{scripts,assets}
# 复制所有文件
cp SKILL.md ~/.claude/skills/easy-read/
cp scripts/generate_glossary.py ~/.claude/skills/easy-read/scripts/
cp assets/glossary_template.html ~/.claude/skills/easy-read/assets/然后在 Claude Code 中输入 /easy-read 使用。
/easy-read
> "这篇文章看不懂" [粘贴文章内容]
Skill 会：
- 提取并分析内容
- 识别所有技术术语和行话
- 为每个术语生成大白话解释
- 创建美观的 HTML 词汇表，包含：
- 难度等级（入门级 → 大师级）
- 发音指南（国际音标 + 中式发音）
- 历史背景（术语如何演变）
- 300 字文章摘要
- 保存到桌面的
claudecode/easy-read-output/文件夹 - 在浏览器中打开
纯文本：
/easy-read
> "帮我理解这个：[粘贴文章]"
文件：
/easy-read
> "解释这个文件：~/Documents/ai-article.pdf"
网址：
/easy-read
> "这篇文章看不懂：https://example.com/article"
截图：
/easy-read
> "看不明白这个" [附加截图]
Skill 识别那些会让零计算机基础的人困惑的术语：
- Prompt Engineering（提示词工程）、Context Engineering（上下文工程）、Harness Engineering（驾驭工程）
- Agent（智能体）、LLM（大语言模型）、RAG（检索增强生成）、Fine-tuning（微调）
- Embedding（嵌入）、Token（令牌）、Context Window（上下文窗口）
- API（应用程序接口）、SDK（软件开发工具包）、CLI（命令行界面）、Framework（框架）
- Backend（后端）、Frontend（前端）、Full-stack（全栈）
- Open Source（开源）、Repository（仓库）、Deployment（部署）
- GPT、AI、ML、NLP
- SaaS、PaaS、IaaS
- 大模型、提示词、上下文
- 游戏引用（Dota、英雄联盟）
- 梗和网络文化
- 历史科技事件
每个词汇表包含：
一个 300 字的大白话摘要，捕捉关键点，不假设任何技术知识。
入门级 — 日常生活中可能遇到的基础概念
- 例如："API" → "应用程序接口 — 就像餐厅的菜单，让不同软件之间能互相'点菜'交流"
进阶级 — 需要一些背景知识
- 例如："Prompt Engineering" → "提示词工程 — 学会怎么跟 AI 说话，让它更懂你的意思"
专业级 — 行业从业者常用
- 例如："Context Window" → "上下文窗口 — AI 一次能'记住'多少内容的限制"
专家级 — 深度技术概念
- 例如："Embedding" → "嵌入向量 — 把文字转成数字，让 AI 能理解和比较"
大师级 — 前沿研究
- 例如："Harness Engineering" → "驾驭工程 — 设计 AI 的工作环境和工具，而不是写更长的提示词"
- 中文名称 — 中文翻译
- 英文名称 — 原始术语
- 发音 — 国际音标 + 中式发音提示
- 大白话解释 — 用日常语言解释含义
- 历史背景 — 术语如何演变（追溯 1-2 个版本）
- 时间线 — 何时流行起来
这个 skill 使用渐进式工作流：
| 文件 | 用途 | 何时使用 |
|---|---|---|
SKILL.md |
核心工作流程和规则 | 始终（skill 调用时） |
scripts/generate_glossary.py |
HTML 生成逻辑 | 阶段 3（输出创建） |
assets/glossary_template.html |
视觉模板 | 阶段 3（输出创建） |
设计遵循原则："让复杂的东西变得易懂，而不是简化。"
这个 skill 诞生于以下信念：
-
信息应该对每个人都可及。 你的奶奶应该能够阅读 AI 新闻而不感到迷失。
-
术语是障碍，不是特色。 技术术语存在是为了精确，但它们不应该把人们排除在理解重要发展之外。
-
背景很重要。 知道一个术语从哪里来，有助于你理解它要去哪里。
-
美观的设计有助于理解。 设计良好的词汇表不仅仅是好看 — 它帮助你学习。
- Claude Code CLI
- Python 3.7+（用于词汇表生成）
- 处理网页内容需要：
web-accessskill 或类似工具
当你处理一篇关于"Harness Engineering"的文章时，你会得到一个词汇表，解释：
- Harness Engineering（大师级）— "驾驭工程：不是教 AI 怎么做，而是给它设计好工作环境和工具"
- Prompt Engineering（进阶级）— "提示词工程：学会跟 AI 说话的艺术"
- Agent（专业级）— "智能体：能自己做决策、调用工具的 AI 程序"
- Context Window（专家级）— "上下文窗口：AI 一次能'看'多少内容"
加上一个 300 字的摘要，用大白话解释文章的要点。
Easy Read helps non-technical people understand complex technical content. It automatically identifies difficult terms, creates beautiful glossaries with plain-language explanations, and provides article summaries that make sense to complete beginners.
Here's what makes it special:
https://github.com/user-attachments/assets/[demo-video-placeholder]
- Zero Technical Background Required — Designed for grandparents, students, and anyone curious about tech news
- Smart Jargon Detection — Automatically identifies terms that need explanation (AI concepts, acronyms, buzzwords)
- Beautiful Output — Clean, modern HTML glossaries with difficulty levels, pronunciation guides, and historical context
- Multiple Input Formats — Accepts text, files, URLs, PDFs, screenshots — anything you throw at it
- Information Equity — Everyone deserves to understand what's happening in tech, regardless of background
Install directly from Claude Code:
/plugin marketplace add HeiGeAi/easy-read
/plugin install easy-read@easy-readThen use it by typing /easy-read in Claude Code.
Copy the skill files to your Claude Code skills directory:
# Clone the repository
git clone https://github.com/HeiGeAi/easy-read.git ~/.claude/skills/easy-readOr manually copy:
# Create the skill directory
mkdir -p ~/.claude/skills/easy-read/{scripts,assets}
# Copy all files
cp SKILL.md ~/.claude/skills/easy-read/
cp scripts/generate_glossary.py ~/.claude/skills/easy-read/scripts/
cp assets/glossary_template.html ~/.claude/skills/easy-read/assets/Then use it by typing /easy-read in Claude Code.
/easy-read
> "I can't understand this article" [paste article text]
The skill will:
- Extract and analyze the content
- Identify all technical terms and jargon
- Generate plain-language explanations for each term
- Create a beautiful HTML glossary with:
- Difficulty levels (Entry → Master)
- Pronunciation guides (IPA + Chinese phonetics)
- Historical context (how the term evolved)
- A 300-word article summary
- Save to your Desktop in
claudecode/easy-read-output/ - Open it in your browser
Text:
/easy-read
> "Help me understand this: [paste article]"
Files:
/easy-read
> "Explain this file: ~/Documents/ai-article.pdf"
URLs:
/easy-read
> "I can't understand this article: https://example.com/article"
Screenshots:
/easy-read
> "Can't understand this" [attach screenshot]
The skill identifies terms that would confuse someone with zero computer science background:
- Prompt Engineering, Context Engineering, Harness Engineering
- Agent, LLM, RAG, Fine-tuning
- Embedding, Token, Context Window
- API, SDK, CLI, Framework
- Backend, Frontend, Full-stack
- Open Source, Repository, Deployment
- GPT, AI, ML, NLP
- SaaS, PaaS, IaaS
- 大模型, 提示词, 上下文
- Game references (Dota, League of Legends)
- Memes and internet culture
- Historical tech events
Each glossary includes:
A 300-word plain-language summary that captures the key points without assuming any technical knowledge.
Entry Level — Basic concepts you might encounter in daily life
- Example: "API" → "Application Programming Interface — like a restaurant menu that lets different software 'order' from each other"
Intermediate — Requires some context
- Example: "Prompt Engineering" → "The art of learning how to talk to AI so it understands you better"
Professional — Industry practitioners use these
- Example: "Context Window" → "The limit on how much content AI can 'remember' at once"
Expert — Deep technical concepts
- Example: "Embedding" → "Converting text into numbers so AI can understand and compare"
Master — Cutting-edge research
- Example: "Harness Engineering" → "Designing AI's work environment and tools, rather than writing longer prompts"
- Chinese Name — 中文名称
- English Name — Original term
- Pronunciation — IPA phonetic + Chinese pronunciation guide
- Plain Explanation — What it means in everyday language
- Historical Context — How the term evolved (1-2 versions back)
- Timeline — When it became popular
This skill uses a progressive workflow:
| File | Purpose | When Used |
|---|---|---|
SKILL.md |
Core workflow and rules | Always (skill invocation) |
scripts/generate_glossary.py |
HTML generation logic | Phase 3 (output creation) |
assets/glossary_template.html |
Visual template | Phase 3 (output creation) |
The design follows the principle: "Make complex things accessible, not simplified."
This skill was born from the belief that:
-
Information should be accessible to everyone. Your grandmother should be able to read AI news without feeling lost.
-
Jargon is a barrier, not a feature. Technical terms exist for precision, but they shouldn't exclude people from understanding important developments.
-
Context matters. Knowing where a term came from helps you understand where it's going.
-
Beautiful design aids comprehension. A well-designed glossary isn't just pretty — it helps you learn.
- Claude Code CLI
- Python 3.7+ (for glossary generation)
- For web content:
web-accessskill or similar
When you process an article about "Harness Engineering", you'll get a glossary that explains:
- Harness Engineering (Master) — "Designing AI's work environment and tools, not teaching it what to do"
- Prompt Engineering (Intermediate) — "The art of learning how to talk to AI"
- Agent (Professional) — "An AI program that can make decisions and use tools on its own"
- Context Window (Expert) — "How much content AI can 'see' at once"
Plus a 300-word summary explaining the article's main points in plain language.
Created by @blakexu with Claude Code.
Inspired by the belief that everyone deserves to understand the technology shaping our world.
受信念启发：每个人都应该理解正在塑造我们世界的技术。
MIT — Use it, modify it, share it. | 使用它，修改它，分享它。