4-Platform Blog Auto-Publishing: Fully Automated Pipeline

Source: Dev.to

4‑Platform Blog Auto‑Publishing: A Fully Automated Pipeline from Writing to Live

Writing blogs isn’t hard. Publishing to multiple platforms is. Copy‑paste, format adjustments, image uploads — exactly the repetitive work AI agents excel at eliminating.

Platform Selection

Rejected: Medium (API essentially abandoned), Note.com (zero API). If you can’t publish via API, it’s not worth including in automation.

Zenn: GitHub Sync

Place Markdown files in articles/, push to GitHub → Zenn auto‑syncs. Version control is naturally integrated, making rollbacks easy.

Token Persistence Lesson

Initially stored tokens in the agent’s session context. When the session ended or the agent restarted, the tokens disappeared. Any state needed across sessions must be persisted to files or a database — never rely on context memory.

Bilingual Publishing Flow

1. Write in Chinese.
2. Create Japanese version (Zenn + Qiita).
3. Create English version (Hashnode + dev.to).

Translation isn’t just language conversion; it requires cultural adaptation. The Japanese version uses politer expressions, while technical terms are kept in English (a common convention in the Japanese tech community).

Publishing Pipeline

1. Write (Chinese Markdown)
2. Translate (JA + EN)
3. Format adaptation
4. Parallel publish to the four platforms
5. Verify
6. Record

From completion to all platforms live: usually under 5 minutes.

Reflections

An AI agent’s capability boundary often isn’t “can it do this?” but “is it worth automating?” If a platform doesn’t provide an API, move on — there are plenty of platforms that embrace automation.