stripe
Fail
Audited by Gen Agent Trust Hub on Feb 14, 2026
Risk Level: HIGHPROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [Indirect Prompt Injection] (HIGH): The skill possesses high-privilege capabilities while ingesting untrusted data from an external API. \n
- Ingestion points: Data returned from
api.stripe.com(e.g., customer metadata, invoice descriptions). \n - Boundary markers: None identified; external data is processed as raw text within the agent's context. \n
- Capability inventory: Ability to perform refunds, create payment intents, and manage subscriptions via
curl. \n - Sanitization: No logic is present to sanitize or validate data retrieved from the API before it influences agent behavior. \n- [Data Exposure & Exfiltration] (HIGH): The skill requires the
STRIPE_API_KEYenvironment variable. The documentation encourages the use of live secret keys (sk_live_), which grant full administrative access to the associated Stripe account and financial data. \n- [Command Execution] (MEDIUM): The skill usescurlandjqto interact with the Stripe API. If user-provided parameters or data from the API are not strictly escaped when interpolated into these shell commands, it creates a risk of command injection.
Recommendations
- AI detected serious security threats
Audit Metadata