The Asset Administration Shell (AAS) serves as an interoperable digital representation of assets using Submodels to encapsulate domain-specific aspects. Its Value-Only JSON representation omits semantic identifiers, data types, cardinalities, and qualifier constraints, eliminating the possibility of structural validation based on the AAS metamodel alone. This paper presents an automated approach that generates JSON Schemas from Submodel Templates (SMTs) to restore validation capabilities for Value-Only payloads. The method constructs an intermediate representation that captures SMT structure and common qualifier types and systematically translates them into executable JSON Schema constructs. A prototype implementation processes current IDTA templates within milliseconds and produces schemas suitable for off-the-shelf JSON validators. Evaluation on representative SMTs demonstrates that both structural and qualifier constraints are preserved in the generated schemas, enabling early detection of modeling and implementation errors. The approach integrates seamlessly into existing AAS API ecosystems and enhances the reliability of JSON-based data exchange.