Module 9 Exercises — Ship a custom skill + subagent you'll actually reuse
Now you bend Claude Code to YOUR way of working. The skill checks lock in the key files and folders; the real-work challenges produce a CLAUDE.md and a skill you'll keep using.
Quick skill checks
One right answer each — paste Claude's output and grade yourself instantly.
The file Claude reads first
Which file does Claude Code automatically read at the root of a project to learn your conventions and context? Type the exact filename.
Where slash commands live
Project-specific slash commands are markdown files in a particular folder. What's that folder path (relative to your project root)?
A skill's required file
Every Agent Skill needs one specific definition file (with frontmatter and instructions). What's its exact filename?
Take it to your real work
No single right answer — this is where the value is. Use your own material.
Write a CLAUDE.md for a real project
In a project you actually use, run /init or write CLAUDE.md by hand: what the project is, key commands, conventions, and your preferences. Keep it tight. Start a fresh session and confirm Claude follows it.
Win: A new Claude session 'just knows' your project's rules without you repeating them.
Build a skill you'll reuse
Pick a task you do repeatedly (e.g. 'draft my weekly update from this week's commits'). Create a SKILL.md with a clear name and description, write the instructions, and test that Claude invokes it correctly.
Win: You triggered your own skill and it did the repeatable task the way you would.