idea-validation

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). [1] The required runtime workflow ingests the user’s outsider-authored free text “Idea” at runtime to generate market research/PRD/build plans and also instructs the Researcher subagent to extract “recurring pain points” by reusing users’ actual words verbatim from community search results (e.g., Reddit search/threads) without narrowing to first-party trusted content.

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

• Direct money access detected (high risk: 1.00). The skill explicitly instructs building payment integrations and names payment gateway tooling (e.g., "Stripe", "Laravel Cashier") and checkout/webhook flows. Those are specific payment APIs/operations (checkout, subscription webhooks) that enable sending/handling real transactions — i.e., direct financial execution capability.