modul 4 — scaling yourself
scheduled work: the agent without you
Some work recurs on a clock, and none of it needs you at the keyboard: the nightly check, the dependency alert, the daily report. Schedule the agent and read the results.
~2 min read · ~10 min with the exercise · v1.0 · verified: 2026-07-15
the work that recurs
A slice of your week repeats on a clock: check the deploy, scan the dependencies, review yesterday's errors, verify the docs still match the code. Each item is a well-specified request of the kind module 1 taught you to write. Requests that need no fresh judgment per run are requests you can schedule.
The docs lay out three tiers by where the work runs and what it can reach. In a session, /loop re-runs a prompt on an interval; it lives and dies with the conversation, which suits polling a build you're waiting on. On your machine, desktop scheduled tasks run with access to your local files. And off your machine, routines (a saved Claude Code configuration: a prompt, repositories, and connectors that run on a schedule, an API call, or a GitHub event, on cloud infrastructure) run "on Anthropic-managed cloud infrastructure, so they keep working when your laptop is closed" (docs).