building-python-web-apps

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

• Third-party content exposure detected (medium risk: 0.30). The skill templates describe an app that processes outsider-authored free text at runtime via HTTP request bodies (e.g., user-submitted payloads to routes like the “/scan” handler and Stripe webhook raw body), which are the inputs an LLM feature (agent.run(prompt)) would consume if wired into those handlers.

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

• Direct money access detected (high risk: 1.00). The skill explicitly includes payment integration (Stripe ≥15), a dedicated services/stripe module for SDK calls, checkout/subscriptions/webhook handling, idempotency, and Terraform provisioning for Stripe. These are specific, purpose-built payment components (payment gateway integration) that enable moving money, so this skill grants direct financial execution capability.