Markdown
Markdown is a lightweight plain-text markup language designed by John Gruber in 2004. It uses concise symbolic syntax—such as hash signs, asterisks, and hyphens—to annotate text formatting, which tools can then convert into HTML and other rich-text formats. Markdown is widely used in documentation, technical blogs, content management systems, and code-hosting platforms, and is characterized by its readability in both source and rendered forms.
Core Facts
skill.md is written in Markdown format and contains metadata (skill name and description) as well as detailed execution instructions.
SDD（Specification-Driven Development，规范驱动开发）是当前AI编程领域被普遍认可的一种开发思想，核心理念是基于规范驱动进行编程
Markdown is a lightweight markup language created by John Gruber in 2004.
Timeline (last 90 days)
Markdown is currently the most suitable intermediate format for LLMs and RAG to process PDFs, because it preserves structures such as headings, lists, and tables while being far smaller in size than the original PDF.
Feeding the messy text extracted from a raw PDF to an LLM wastes the context window and introduces hallucination triggers, whereas clean Markdown tables and heading levels can improve vector retrieval accuracy.
MD-Convert's PDF-to-Markdown tool's processing flow is: raw PDF specification -> semantic AST parsing -> clean Markdown file, and the output can be used directly for GitHub documentation, LLM context windows, and static site generators.
When processing Markdown or EPUB-format e-books, it is necessary to first perform structured parsing and remove non-readable content such as Markdown heading symbols, footnote markers, and image links, otherwise problems such as reading heading symbols aloud and footnotes being mixed into the body text will occur.
AIClaw's local Markdown preview prohibits image tags and inline style attributes, embedded images are not automatically fetched, and relative file references are resolved from the document's directory; it does not execute MDX components and is not a browser for arbitrary HTML applications.
Andrej Karpathy proposed the LLM Wiki pattern, which uses Markdown pages plus YAML metadata to organize knowledge, allowing humans and AI agents to maintain it together and query it on demand.
MarkAny is a tool that can batch-convert formats such as PDF, Word, Excel, PowerPoint, HTML, and EPUB into structured Markdown and supports reverse export back to common formats, while also featuring built-in OCR capabilities.
Exporting to external-facing formats from cleaned Markdown source files is preferable to directly editing the original binary files, in order to avoid version confusion across multiple copies.
Re-rendering Markdown on every token causes long responses to become progressively slower; Markdown should be rendered all at once when the done event is received, or at most once every approximately 100 milliseconds.
AI chat interfaces experience rendering flickering during streaming output because Markdown parsers assume they are receiving a complete document, whereas in a streaming scenario only a prefix fragment is available.
40 more timeline events
All Facts (20)
skill.md is written in Markdown format and contains metadata (skill name and description) as well as detailed execution instructions.
80%VerifiedSDD（Specification-Driven Development，规范驱动开发）是当前AI编程领域被普遍认可的一种开发思想，核心理念是基于规范驱动进行编程
80%VerifiedMarkdown is a lightweight markup language created by John Gruber in 2004.
80%VerifiedMarkdown is both human-readable and easy for large language models to parse, meeting the needs of both human authoring and machine reading.
75%VerifiedMarkdown is a lightweight markup language with concise syntax, strong readability, and ease of version management, and has become the de facto standard format for technical documentation.
70%VerifiedLarge language models encountered a large volume of Markdown-formatted text from sources such as GitHub, technical documentation, and Wikipedia during the pre-training phase, and therefore have an extremely strong understanding of Markdown structure.
70%VerifiedMarkdown is the default document format for mainstream platforms such as GitHub, Notion, and Obsidian, and is also the default output format for nearly all large language models.
70%VerifiedSkills files typically contain YAML front matter (for trigger condition evaluation) and Markdown body text (for defining workflows and output specifications).
65%VerifiedOpenAI's GPT series models are particularly friendly to Markdown format, because they were extensively exposed to code and documentation on GitHub—primarily in Markdown format—during training.
65%VerifiedHTML output allows AI to embed SVG charts, interactive components, in-page navigation, color coding, and inline annotations—features that Markdown cannot achieve.
65%UnverifiedMarkdown是GitHub、Stack Overflow、Notion等平台的标准格式
95%UnverifiedMarkdown作为纯文本格式与Git的diff机制完美兼容，而HTML文件由于包含大量标签和样式代码，使得代码审查变得极其困难
90%UnverifiedHTML输出比Markdown消耗明显更多的Token，一个简单的加粗文字在Markdown中需要6个额外字符，而在HTML中需要17个额外字符
90%UnverifiedHTML的diff极其嘈杂难以审查，是HTML替代Markdown的最大缺点
85%Partially VerifiedObsidian是一款基于本地Markdown文件的双向链接笔记工具，所有数据以纯文本形式存储在本地
80%UnverifiedMarkdown的结构化标题层级天然对应需求文档的章节层次，且对大模型友好，token利用率高于HTML等富格式文本
80%UnverifiedMarkdown适合技术文档、笔记、知识库等记录和传承场景，HTML适合报告、看板、演示、原型等理解和行动场景
70%UnverifiedAI模型处理Markdown文件的效率远高于PDF
70%UnverifiedMarkdown has become a universal information carrier in the AI era due to its structured plain text and low-redundancy characteristics.
60%UnverifiedMainstream LLMs are exposed to large amounts of Markdown-formatted documents in their training data, and therefore their ability to parse and follow Markdown structure is superior to other formats.
60%