referral-program

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly instructs the agent at runtime to read and "use" the file ".claude/product-marketing-context.md" before asking questions (SKILL.md:8), meaning fetched content would directly influence the agent's prompts and behavior.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly references payment gateway integrations used to handle commissions/rewards — e.g., "Rewardful" (Stripe-native) and "Stripe | Payment processing (for commission tracking)". These are specific payment APIs/tools that enable moving money/handling payouts, so the skill grants direct financial execution capability.