Construct and blueprint
Assessment must identify the capability/construct, population, criteria, evidence channel and intended use.
Published versions
Published assessment versions/rubrics are immutable for the attempt they govern; later changes create new versions rather than silently changing historical criteria.
Automatic scoring
Automatic finalization is limited to permitted objective/formative cases. Complex or consequential judgments require the human-review rules defined by the assessment engine.
AI
AI may assist within explicit bounds but cannot publish, moderate or own consequential assessment decisions where human authority is required.
Interpretation
Assessment does not automatically create mastery, readiness, credential or verification truth; those are separately governed states.
Accessibility
Accommodations should preserve the intended construct while enabling equitable access.
Pre-empirical status
Where empirical reliability/validity has not been established, Certifiedge must say so rather than invent statistics.
Questions or requests
Use the Certifiedge contact, complaints or data-rights route appropriate to the issue. Internal policies do not remove rights or remedies available under applicable law.