JSON from a model is passed to a business workflow.
Prevent unsafe values and fields.
Choose two controls.
Validate schema, types, required fields, ranges, and business rules, and reject or review invalid output before side effects.
Detailed explanation
Downstream assumptions are protected.
Syntax and meaning differ.
Injection and mistakes can execute.
Meaning and safety are verified.
Check official documentation
Compare the following points with the official documentation.
Review Task 5.1 and secure output handling. For hypothetical schema-compliant JSON, separate checks for the target record, permitted action, current business state, and side effects, without allowing model-supplied values to expand permissions.
Official references
What to confirm
Separate structural conformance from business validity and authorization, checking trusted permissions and current state before execution.
Key points
- JSON Schema
- Range
- Side effect
Notes
- Environment: Reference review of official documentation and hypothetical examples; no AWS API calls, business data access, or downstream execution are required.
- Do not use a generated user ID or permission flag as a substitute for trusted authentication and authorization. Check the target and current permissions in the application.
- Schema conformance, factual accuracy, and business permission are separate checks. Parameterized SQL and encoding for the display context remain necessary after schema validation.
Foundation review
Semantic and business validation
A correctly typed value can still violate a record state or business rule. Check those conditions separately from structure.
Authorization and side effects
Authorize actions using the trusted principal and inspect changes such as writes or sends. A permission name generated by the model is not an authorization decision.