Module 10 — Putting the pieces together: a worked example

Theory is easier to hold when you see the parts cooperate. Here’s a realistic Cowork-style task and which features fire, so the whole machine clicks.

“Every Monday, gather last week’s support tickets, summarize the top complaints, and put them in a Word doc in my Reports folder.”

What happens, mapped to the modules

Connector (Module 9): Claude uses the support-tool connector to pull last week’s tickets. The raw tickets are bulky, so it doesn’t keep all of them on the main desk.
Subagent (Module 8): It might hand the “read 200 tickets and extract complaints” job to a subagent, which churns through them on its own desk and returns a short list of themes — keeping the main context (Module 4) clean.
Skill (Module 5): To produce a proper .docx, Claude matches the task against a document-creation skill, loads it, and follows its instructions (and runs its scripts) to build a real Word file — not a guess.
Memory / CLAUDE.md (Module 3): Your standing instructions (“our reports always start with an executive summary, use the company logo”) shape the format automatically.
Hook (Module 7), optional: a PostToolUse hook could copy the finished file to a backup folder every time.
Scheduled task (Module 11): the “every Monday” part means this whole thing runs on a schedule, untouched by you.

No single feature does the job. The art of mastery is knowing which piece handles which part.