prava-sdk-integration
Audited by Snyk on May 17, 2026
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to ask users for their Secret Key and shows examples embedding that key (e.g., Authorization: Bearer {MERCHANT_SECRET_KEY}, client-side SECRET_KEY in sample JS), which would require the LLM to handle and potentially output secret values verbatim.
HIGH W008: Secret detected in skill content (API keys, tokens, passwords).
- Secret detected (high risk: 1.00). I flagged the documented secret because a high-entropy secret-format key is present verbatim. The table under "Required Inputs" contains a Secret Key example: sk_test_zGzBj2QzZVaFtO4dkY2ZLAGe7wRSf1zgzUPBheBksA4
This is not a placeholder like "sk_test_YOUR_KEY" or "sk-xxxx" and looks like a real test-style secret (high entropy, full-length). Therefore it meets the definition to be treated as a leaked credential.
Notes on ignored items: I ignored obvious placeholders and low-security examples elsewhere (e.g., MERCHANT_SECRET_KEY=sk_test_YOUR_SECRET_KEY_HERE, PUBLISHABLE_KEY='pk_test_YOUR_KEY', and other "YOUR_*" values). The publishable key example (pk_test_...) is present too; publishable keys are generally safe for client usage and do not grant secret server-side access, so the primary concern is the sk_test_ secret shown.
MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).
- Direct money access detected (high risk: 1.00). The skill is explicitly a payment integration for a payment gateway (Prava). It includes concrete API calls and SDK methods for creating payment sessions (POST /v1/sessions with a MERCHANT_SECRET_KEY), polling payment results (GET /v1/sessions/{id}/payment-result authenticated with the secret key), and obtaining one-time transaction credentials (Visa network token + dynamic CVV) that are intended to be used to transact. It instructs storing and using publishable/secret keys, shows server-side routes that call the payment backend, and describes using the returned token/CVV to perform transactions. This is a specific financial-execution capability (payment gateway integration), not a generic tool, so it meets the Direct Financial Execution criteria.
Issues (3)
Insecure credential handling detected in skill instructions.
Secret detected in skill content (API keys, tokens, passwords).
Direct money access capability detected (payment gateways, crypto, banking).