How well are the tests covering the code?

A walkthrough of code coverage levels using a Ruby example, progressing from line/statement coverage through branch coverage, path coverage, and condition coverage. Each level is demonstrated with Minitest examples using the Simplecov gem, showing how each successive level catches bugs that the previous level missed. The post also highlights a less obvious insight: if you can't write a meaningful test to cover a piece of code, that code may not need to exist — especially relevant when AI agents generate code without the friction of manual writing.