Choosing Tags for Technical Blog Posts: A Small Taxonomy

Tags are deceptively important for technical writing. A good tag set is small, mutually exclusive at the top level, and stable across years. A bad one grows without limit and stops meaning anything.

YJTOON's top-level tags

• toon — anything about the format itself.
• llm — model behavior, prompting, evaluation.
• api — YJTOON endpoints, rate limits, error shapes.
• data — dataset design, schema, provenance.
• tokens — token counts, budgets, encoders.
• agents — agentic patterns that consume YJTOON.

When to add a tag

Only when you have three posts that need it. Two posts is a coincidence. One post is a single topic, not a tag.

When to retire a tag

When nothing has been tagged with it for a year, and nothing in the backlog wants it. Retired tags are kept as redirects, not deleted.

How this helps models

Tags are visible in the index, the RSS feed, and the per-post meta. A model scanning the index can filter by tag without parsing the body. That is the whole point.