autumn

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

• Third-party content exposure detected (high risk: 0.95). The required workflow fetches public documentation at runtime from https://docs.useautumn.com/llms.txt and then fetches additional markdown pages via WebFetch, so outsider-authored free text from the public docs is ingested into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs a runtime sub-agent to fetch external documentation starting with https://docs.useautumn.com/llms.txt (and related https://docs.useautumn.com/{path}.md pages) and inject that content into the agent's prompt/context to drive the research, so these URLs are runtime dependencies that directly control agent behavior.

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

• Direct money access detected (high risk: 1.00). The Autumn skill is explicitly a billing/monetization platform that "integrates with Stripe under the hood" and exposes billing-specific APIs. The documentation lists concrete endpoints and features for creating/updating subscriptions and payment sessions (e.g., /api-reference/billing/attach.md, /api-reference/billing/billingUpdate.md, /api-reference/billing/setupPayment.md, /api-reference/billing/createSchedule.md, openCustomerPortal.md), platform endpoints for Stripe OAuth and updating connected Stripe accounts (/api-reference/platform/oauth-url.md, /api-reference/platform/update-stripe.md), and billing behaviors like auto top-ups and webhooks (billingAutoTopupSucceeded). These are specific payment gateway and subscription management capabilities (not generic tools), and therefore grant direct financial execution authority.