recoup-internal-weekly-usage-review

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt explicitly requires use of multiple real API keys/tokens (Privy, Stripe, AI_GATEWAY, a fresh Vercel session token) and even recommends passing a secret on the command line (--api-key "$STRIPE_RESTRICTED_KEY") or reading a session token file, which encourages the agent to include or handle secret values verbatim in commands/requests.

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

• Direct money access detected (high risk: 1.00). The skill explicitly requires and documents use of Stripe (a payment gateway): it asks for an authenticated Stripe CLI or live restricted key, references Stripe CLI commands and Stripe API endpoints (subscriptions, invoices, charges, events, /v1/payouts), and inspects billing/payout/subscription data. Those are specific payment-gateway integrations (not generic HTTP or browser tooling), so this skill exposes direct financial execution authority risk (presence of Stripe credentials/APIs). Note: the skill states calls are read-only, but it still requires/uses Stripe keys and API paths, which matches the criteria for payment-gateway capability.