Taiwan.md Knowledge Base Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection. Taiwan.md is an open source, AI native knowledge base about Taiwan built with Astro v5. It uses a Single Source of Truth (SSOT) architecture where all content lives in the knowledge/ directory as Markdown files, and the website is a build time projection. Features include bilingual support (Traditional Chinese as default + English), an interactive D3.js knowledge graph, and 96+ curated articles across 12 categories. Installation & Setup Prerequisites Node.js 18+ npm or pnpm Clone and Install Development Server Build & Preview Sync Knowledge to Content Project Architecture Critical rule: Never edit files in src/content/ directly. Always edit knowledge/ and run scripts/sync.sh . Content Structure The 12 Categories Slug Chinese English history 歷史 History geography 地理 Geography culture 文化 Culture food 美食 Food art 藝術 Art music 音樂 Music technology 科技 Technology nature 自然 Nature people 人物 People society 社會 Society economy 經濟 Economy lifestyle 生活 Lifestyle Article File Naming Writing Articles Chinese Article Template ( knowledge/[Category]/article slug.md ) English Article Template ( knowledge/en/[Category]/article slug.md ) Hub Page Template ( knowledge/[Category]/ Hub.md ) Frontmatter Reference Required Fields Optional Fields Adding Images (Wikimedia Commons Policy) All images must be from Wikimedia Commons with verified CC licenses. Cache them locally: Image attribution format in frontmatter: Knowledge Graph Integration Articles automatically appear in the D3.js knowledge graph at /graph . Nodes are created from articles; edges are created from shared tags and cross references. Linking Articles Reference other articles within content using relative paths: Tagging for Graph Connections Use consistent tags to create knowledge graph bridges: Sync Workflow After editing any file in knowledge/ , always sync before building: Three Layer Depth Pattern Every article should follow this structure for AI readability and varied reading levels: Contributing via PR Full PR Workflow Commit Message Convention AI Native Features llms.txt The site exposes /llms.txt for AI consumption. When writing content, use structured headers that AI can parse: Meta AI Summary Tag Pages include <meta ai summary — write descriptions that work as standalone AI context: Astro Page Patterns Category Page ( src/pages/[category].astro ) Bilingual Route Pattern Content Quality Checklist Before submitting a PR, verify: [ ] Both knowledge/[Category]/article.md (zh TW) and knowledge/en/[Category]/article.md (en) exist [ ] All sources entries have clickable url fields (no plain text only references) [ ] Article follows three layer depth: 30 sec → deep dive → why it matters [ ] Images sourced from Wikimedia Commons with proper imageCaption attribution [ ] category slug matches one of the 12 valid categories exactly [ ] bash scripts/sync.sh && npm run build completes without errors [ ] Factual claims are verified against cited sources Troubleshooting Build fails after adding article Article not appearing in knowledge graph English article not showing at /en/... Images not loading Sync script permission error Resources Live site: https://taiwan.md Knowledge graph: https://taiwan.md/graph Contribute guide: https://taiwan.md/contribute llms.txt: https://taiwan.md/llms.txt License: CC BY SA 4.0 (content) + MIT (code) Contact: cheyu.wu@monoame.com

taiwan-md-knowledge-base