Python developers use decorators and descriptors on a daily basis, but many don't understand them well enough to create (or debug) them. Decorators are widely deployed in popular Python Web frameworks. Descriptors are the key to the database mappers used with those frameworks, but under the covers they play an even more crucial role in Python as the device that turns plain functions into bound methods, setting the value of the `self` argument. This tutorial is a gentle introduction these important language features, using a test-driven presentation and exercises, and covering enhancements in Python 3.6 that make class metaprogramming easier to get right. Decorators without closures are presented first, highlighting the difference between _run time_ and _import time_ that is crucial when meta-programming. We then get a firm grounding on closures and how they are implemented in Python, before moving to higher order function decorators and class decorators. Coverage of descriptors starts with a close look at Python's `property` built-in function and dynamic attribute look up. We then implement some ORM-like field validation descriptors, encounter a usability problem, and leverage PEP 487 -- Simpler customisation of class creation -- to solve it. Alternative implementations using a class decorator and a metaclass will be contrasted to the PEP 487 solution.