Notes
Technical thinking, not content marketing cosplay.
Short essays about architecture, AI systems, publishing infrastructure, and the weird places where software complexity quietly mutates into organizational complexity.
Technical note
publishedHow we reduced Mermaid rendering from three seconds to under 200 ms
A small architectural change: instead of starting Mermaid CLI and a fresh Chromium instance for every diagram, keep one warm browser process alive and call mermaid.render directly.
Open note →
Strategic note
publishedAI as an execution layer: why one builder can now operate like a small team
AI is not only a faster way to write code or content. It changes the distance between an idea and a working iteration.
Open note →
Systems note
publishedVibe coding as an architectural feedback loop
How an AI agent can become a sensor for architectural friction, not just a faster autocomplete, but a collaborator for context, decisions, and safe execution.
Open note →
Technical note
publishedWhy a Litepress Markdown extension has two parts
On separating editor UX, SSR, validation, preview styles, and AI help in semantic Markdown extensions.
Open note →
Technical note
publishedMarkdown as a semantic runtime, not a text format
Why a Markdown block in Litepress is not just an HTML shortcut, but a structured signal for the renderer, editor, integrations, consent, and AI workflow.
Open note →
Technical note
publishedAI-native authoring: when the editor understands its own language
Why AI in Litepress is not a chat beside the editor, but an MCP-assisted workflow where the agent reads current Markdown instructions based on enabled extensions and integrations.
Open note →
Technical note
publishedWhen the CMS becomes bigger than the problem
Litepress as a technical response to CMS complexity, plugin archaeology, and publishing workflows that become heavier than the content itself.
Open note →