You demonstrate the skill against a situation that shifts under you. Every judgement we publish is tied to a specific thing you did — quoted, timestamped, and checkable.
A live simulation or a real artefact — never multiple choice, and nothing that can be memorised in advance.
The client turns out to have misdiagnosed her problem. The brief changes after you commit. Adapting is the assessment.
Fixed public anchors, scored 0/1/2. No score above zero exists without a quote or a measurement behind it.
A threshold fixed in advance, not a model’s opinion. The credential names the competency, the level and its assurance tier.
The result sits on a public profile you control, so you prove a competency once rather than in every application.
Rubrics and level definitions are published. A hiring manager can judge the system without seeing a single candidate.
A live client who does not know what their real problem is. Uncover it, reframe it, and get them to agree.
Compose a poster from supplied elements. Then the brief changes, and you have to adapt rather than restart.
Levels differ by how much the situation misleads you, not by how hard the task is. Each level states the assurance it can honestly claim — so a verifier knows what weight to give it. V1 verifies at L2.
Two competencies are verifiable end to end right now.