Blog App

apps/blog is a TanStack Start application deployed to Cloudflare Pages. It pre-renders 393 static pages at build time.

Isomorphic data loading

The blog uses readPublicJson() in apps/blog/lib/posts.ts for loading post data:

• During SSR / prerender: reads directly from the filesystem (fs.readFile)
• During client navigation: uses fetch() against the deployed URL

This is unique to the blog. Other apps use absolute external API URLs, which work in both SSR and client contexts without special handling.

Markdown rendering

Markdown is converted to HTML by the Rust-backed WASM crate at packages/wasm/. The exported markdownToHtml function from @duyet/libs delegates to the WASM binary. This gives an 79x speedup (6.3 ms → 0.08 ms) over the pure-TypeScript unified/remark pipeline it replaced.

The WASM binary is gitignored and must be built in CI before the blog build step. Absence of the binary causes silent prerender failures.

Slug normalization

Post frontmatter slugs historically had a .html suffix (Jekyll legacy). These are stripped at data generation time. getPostBySlug() includes defensive normalization to handle both formats.

Test coverage

50 tests as of last count — covers post loading, slug normalization, and rendering utilities.

Design

White #ffffff background (distinct from the warm cream #fbf7f0 used by other apps). Dark mode uses 20% opacity card tints. See blog design article for full conventions.

Copy dropdown

Post pages have a copy dropdown offering:

• Copy (clipboard)
• View as Markdown
• Open in ChatGPT
• Open in Claude

Series support

SeriesBox is rendered on post pages to show series context. It was removed briefly in the March 2026 cleanup but restored on 2026-03-28 after the user confirmed series context is useful.

Plain-text access

Individual .md files are generated alongside each post's HTML for plain-text access (useful for LLM ingestion and llms.txt pipelines).