Testing & Qualification
Welcome to the HardKAS Testing & Qualification documentation.
Because HardKAS is a determinism and evidence-oriented framework, how we prove a feature works is just as important as the feature itself.
This section defines the canonical vocabulary for capability tracking, evidence isolation, and test suite execution.
- Evidence Levels & Qualification Model: Understand the difference between L2 (Simulator) and L3 (Localnet) evidence, and how
IMPLEMENTEDdiffers fromQUALIFIED. - Test Suites & Commands: How to run the tests, where to place new tests, and how HardKAS enforces strict evidence isolation (no silent skipping if Docker is unavailable).
Core Principle
Documentation communicates qualification; it does not manufacture it.
A capability is only marked as L3 if there is a *.localnet.test.ts (or equivalent automated run) proving it works against a real Kaspa node. An anecdotal run by a developer does not qualify as permanent L3 evidence.