Core concept

Playbook

A repeatable recipe that chains several steps into one finished outcome, so a whole job runs the same way every time instead of being pieced together by hand.

The anatomy of a playbook A trigger starts the playbook, a fixed set of steps runs in order, and it ends in one finished output. Input lands 1. Gather 2. Shape 3. Check your eyes One finished thing Same input, same steps, same output, every time.

A playbook is a recipe for work: a fixed set of steps that turns a known input into a known output, the same way every time. The value is that it repeats. A one-off question is a task; "every time a contract lands, pull the parties, the term, and the get-out clauses, then flag anything unusual" is a playbook.

The shift is from asking Claude a question to running Claude through a job. A good playbook names the trigger (what starts it), the steps in order (gather, shape, check), and the single finished thing it produces. It pulls together everything else in this course: projects for context, skills for method, connectors for reach, all running in Cowork.

Build one by spotting a job that repeats, naming the steps out loud, testing it once by hand, and adding a checkpoint on any risky step. When it works, put it on a scheduled task, or share it as a skill so your whole team runs it the same way.

Often confused with: A single prompt. A prompt is one instruction; a playbook is a sequence of steps with a clear input, checkpoints, and one finished output at the end.

See it in a lesson: Putting it together: work playbooks

Further reading

All terms