An executable acceptance test turns part of a definition of done into a repeatable operation. It specifies the command, environment and success evidence needed to judge the result without relying on a narrative status report.
A trustworthy check requires both a successful process exit and the expected decisive output when output is part of the proof. The command itself must also be reviewed when it can change files, access credentials or affect external systems.
ELI5
An executable acceptance test is a command that checks whether a specific requirement has been met. It turns part of the definition of done into repeatable evidence instead of relying only on someone's claim that the work is complete.
For example, a test can open a sample order, apply a discount, and verify the exact total. The test must finish successfully and show the expected result, and reviewers should understand any files or systems the command can change.
Is a zero exit code enough for an executable acceptance test?
Not always. The command may exit successfully without proving the intended outcome, so relevant output or artifact evidence may also be required.
Why should acceptance commands be stored before work begins?
Predefining them makes the completion standard visible and prevents an agent from choosing an easier check after seeing its implementation.
