Verification Test
Jump to navigation
Jump to search
A Verification Test (V-test) is an executable check paired with a requirement or deliverable that returns pass or fail, distinguishing verified completion from declared completion.
- AKA: V-Test.
- Context:
- It can gate stage boundaries: work is complete when its test passes, not when its author asserts completion.
- It can be paired one-to-one with a requirement identifier in a specification, giving each contract clause an executable falsifier.
- It can be distinguished from a general Software Test by its governance role: it verifies that a claimed deliverable exists and behaves as claimed, rather than exploring behavior generally.
- It can counter completion-declaration bias in both human and agent workflows.
- Example(s):
- A release gate requires each advertised capability to pass its verification test — an executable command whose output confirms the artifact exists and behaves as claimed — before the release tag is created.
- An Acceptance Test tied to a contract clause in a formal requirements document.
- Counter-Example(s):
- An exploratory Unit Test suite — verifies code behavior, not deliverable-against-claim.
- A manual review checklist without an executable check — declaration, not verification.
- See: Software Test, Acceptance Test, Software Specification, Requirement, Quality Gate, Test-Driven Development.
References
- Published from the AGET framework; machine-readable concept: w3id.org/aget/vocab#Verification_Test