A publishing workflow for AI-assisted engineering posts

AI tools make it much faster to start a technical article. They do not remove the team’s responsibility for what gets published.

The default Pletor Tech Blog workflow is:

1. The owner writes a short note on the topic, audience, and core message.
2. Codex or Claude Code drafts the article, code examples, and title candidates.
3. The responsible engineer reviews facts, code validity, and security-sensitive wording.
4. The Korean and English versions are linked with the same translationKey.
5. A GitHub PR reviews wording, links, images, and SEO metadata.
6. We check the Cloudflare Pages preview before merging.

This blog also captures what Pletor learns while building Konduo. Konduo is an integrated operations management platform that connects infrastructure status, metric and log evidence, alert response, access control, and operating history into one workflow.

Frontmatter standard

Every post includes at least the following metadata.

title: "Post title"
description: "One-sentence summary for search and social cards"
published: "2026-06-17"
author: "Pletor Engineering"
lang: "en"
translationKey: "stable-post-key"
tags: ["security", "data"]
draft: false

When republishing to external channels, the republished version points its canonical URL to the original article. The source page remains canonical to itself, and external copies point back to it.

Further Reading

For more context connected to this topic, these posts are also worth reading.

• Starting the Pletor Tech Blog - Introduces the operations and product lens behind the blog.
• Why Did Kafka OOM When Memory Was Still Available? - Revisits Kafka broker memory with heap, page cache, and direct buffers in view.
• Why Turn Immutable YAML Logs into Full Objects? - Looks at filtering immutable logs without building full object trees.