Module 10: Automations & Agents
Hands-on

Module 10 Exercises — Ship one automation that runs end-to-end on its own


The capstone: Claude doing work while you're not watching — safely. The skill checks cover the headless and scheduling basics; the real-work challenges build an automation that runs without you.

Quick skill checks

One right answer each — paste Claude's output and grade yourself instantly.

Print modeAuto-graded

Run it headless

Which flag runs Claude Code non-interactively — it takes your prompt, does the work, prints the result, and exits? Type the flag.

Piping resultsAuto-graded

Machine-readable output

To get structured output you can pipe into another program, you add `--output-format ____`. What value makes it machine-readable?

SchedulingAuto-graded

Read a cron schedule

A scheduler runs your job on the cron line `0 9 * * *`. At what hour of the day (24-hour clock) does it run? Enter just the number.

Take it to your real work

No single right answer — this is where the value is. Use your own material.

Unattended runsSelf-check

Schedule a daily headless job

Write a claude -p command that produces something useful daily (e.g. a summary of yesterday's commits, or a check on a URL). Schedule it with cron or a GitHub Actions schedule. Give it only the permissions it needs.

Win: The job ran on schedule without you, and you scoped its permissions deliberately.

Agentic workflowSelf-check

Build a Monitor → Decide → Act → Notify loop

Compose a small end-to-end workflow: gather some data, let Claude decide if action is needed, take the action (or draft it), and notify you (Telegram, email, Slack). Add one guardrail — a dry-run, an allow-list, or a human-approval checkpoint.

Win: The workflow completed the full loop and pinged you — with at least one guardrail keeping it on a leash.