implement-in-app-purchases
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes the IAPProductCatalog.json configuration file, which represents a surface for indirect prompt injection if the file is modified by an untrusted party.
- Ingestion points: The Assets/Resources/IAPProductCatalog.json file is read and modified during product configuration tasks as described in references/codeless-catalog.md.
- Boundary markers: The skill provides detailed validation checklists and schema requirements in the reference documentation to verify data integrity before the agent processes or saves the catalog.
- Capability inventory: The instructions guide the agent to create scripts that call sensitive UnityEngine.Purchasing methods for purchase initiation and confirmation, and handle receipt data for validation.
- Sanitization: Specific guidance in references/codeless-catalog.md details rigorous checks for unique IDs, valid enum indices, and decimal serialization rules to prevent malformed or malicious data from being processed.
Audit Metadata