technical-writing
关于
This Claude Skill applies rigorous technical writing standards to technical prose like code comments and documentation, enforcing active voice, present tense, and precise language. It transforms vague terms into specific metrics and provides tailored rules for different writing types. Use it whenever clarity and precision in technical communication are critical.
快速安装
Claude Code
推荐/plugin add https://github.com/EvanLavender13/audio-jonesgit clone https://github.com/EvanLavender13/audio-jones.git ~/.claude/skills/technical-writing在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
Technical Writing Standards
Apply these standards to ALL technical writing.
Core Rules
Language Mechanics
- Active voice: "The parser validates input" NOT "Input is validated"
- Present tense: "returns null" NOT "will return null"
- 15-20 words per sentence (max 30 before splitting)
- Code references: Use
file_path:line_numberformat
Precision Requirements
Replace vague terms with specifics:
- "fast" → "completes in <100ms"
- "many" → "up to 1000 items"
- "handles" → "validates/transforms/routes/stores" (specify which)
- "manages" → "creates/updates/deletes/tracks" (specify which)
- "processes" → "parses/filters/aggregates/formats" (specify which)
Writing Types
Code Comments
// BAD: Handle the data
// GOOD: Parse JSON response and extract sample values into ring buffer
- Explain WHY, not WHAT (code shows what)
- One line for simple logic, block for complex algorithms
- Reference related code:
// See also: waveform.c:45
Research/Planning Docs
- State findings, not opinions
- Include sources or evidence
- Mark unknowns explicitly: "TBD", "Needs investigation"
- Separate facts from conclusions
Architecture Docs
- WHAT the component does (1 sentence)
- HOW it fits into the system (1 sentence)
- Reference implementation:
src/audio.c:120
Function/API Documentation
Purpose: [1 sentence - what transformation/effect]
Parameters: [type, constraints]
Returns: [type, conditions]
Errors: [what triggers them]
Quality Checklist
Before finalizing:
- Active voice throughout
- Present tense only
- No vague verbs without specifics
- Quantities have units/ranges
- First use of acronyms spelled out
Examples
BAD: "This handles user data and manages sessions."
GOOD: "Validates user credentials against the database and issues JWT tokens with 24-hour expiration."
BAD: "Component processes requests quickly."
GOOD: "Parses incoming HTTP requests and routes to controllers within 50ms p95."
Anti-patterns
Never write:
- "handles various operations"
- "responsible for managing things"
- "processes data as needed"
- "main functionality"
Always specify WHAT operations, WHICH things, WHAT data.
No Duplication
- Don't summarize content that exists elsewhere
- Reference files instead of repeating their contents
- If documentation exists, link to it
Remember: Every word should add understanding. If it doesn't teach something specific, cut it.
GitHub 仓库
相关推荐技能
content-collections
元Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。
llamaindex
元LlamaIndex是一个专门构建RAG应用的开发框架,提供300多种数据连接器用于文档摄取、索引和查询。它具备向量索引、查询引擎和智能代理等核心功能,支持构建文档问答、知识检索和聊天机器人等数据密集型应用。开发者可用它快速搭建连接私有数据与LLM的RAG管道。
cloudflare-turnstile
元这个Skill提供完整的Cloudflare Turnstile集成知识,用于在表单、登录页面和API端点中实现无验证码的机器人防护。它支持React/Next.js/Hono等框架集成,涵盖令牌验证、错误代码调试和端到端测试等场景。通过运行后台不可见挑战,在保持用户体验的同时有效阻止自动化流量和垃圾信息。
canvas-design
元canvas-design技能可根据用户需求生成原创的.png和.pdf格式视觉设计作品。它通过创建独特的设计哲学(包括形式、色彩、构图等元素)来制作海报、艺术品等静态设计。该技能特别适用于需要原创视觉内容的场景,能自动规避版权风险。
