How this blog works

From a Telegram conversation to a published article: research, review on a preview link, one tap to publish.

This blog is written in conversation. I talk with Hermes, my assistant on Telegram, about a topic or a YouTube video. When something is worth keeping, Hermes turns it into an article, and I publish it without leaving the chat.

The publishing pipeline at a glance — Telegram → Hermes → pull request → preview → publish

The pipeline

sequenceDiagram
    actor Me
    participant H as Hermes (Telegram)
    participant G as GitHub
    participant CI as CI checks
    participant V as Vercel
    Me->>H: "Write an article from this"
    H->>G: branch + Markdown + pull request
    G->>CI: lint, schema, links, path guard
    G->>V: preview deployment
    H-->>Me: PR link + preview link
    Me->>H: "publish"
    H->>G: merge (I tap approve)
    G->>V: production deployment

Why Markdown in git

Reviewable: every article is a pull request with a preview link, so nothing goes live unseen.
Safe: articles are plain Markdown. No MDX, no raw HTML, so an article can't run code.
Durable: no database to pause or migrate. The site is static and served from Vercel's CDN.

NOTE

Hermes can only touch apps/blog/content/ and can't change the code or the checks that guard it.