referral-viral-loops
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides logic and instructions for a system that processes untrusted external data, specifically referral codes and customer order data.
- Ingestion points: Untrusted data enters the system through URL parameters and cookies (captured as
referralCode) and order metadata (referenced inSKILL.mdand associated evaluation tasks). - Boundary markers: The instructions do not specify explicit delimiters or security markers to isolate untrusted content from the agent's core instructions when processing these data points.
- Capability inventory: The skill outlines operations for database writes, interaction with e-commerce platforms (Shopify, WooCommerce, BigCommerce), and reward fulfillment mechanisms.
- Sanitization: The implementation examples include basic character stripping (using regex
/[^a-zA-Z]/g) during the generation of referral codes to ensure they remain alphanumeric.
Audit Metadata