Same pattern in content ops — encoding a domain workflow as a skill series

[pencil20388-eng]

Really like how this packages job-hunting into a system of skill modes rather than one monolithic skill. The "one domain, multiple coordinated skills" pattern resonates.

I've been doing the same thing for content operations. Instead of one big skill, I split a content operator's real workflow into three coordinated skills:

• content-topic-judgment — decide whether a topic is worth writing before writing it (relevance, signal, depth, timing)
• stop-slop-zh — strip AI tells from the output (Chinese-focused, but the idea generalizes)
• ai-website-building — turn the content into a deliverable website with no coding

The underlying belief is the same one career-ops seems built on: a domain expert's judgment can be encoded as explicit, AI-runnable rules — and a workflow is better as a coordinated series than one giant prompt.

Curious how you landed on 14 modes as the right granularity. Did you start smaller and split, or map the whole job-search process upfront? Asking because I'm about to package my three into a more explicit system and wondering about the right level of granularity.