agent-agentic-payments
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data within a workflow that includes sensitive payment authorization and transaction signing capabilities.
- Ingestion points: Untrusted data enters the agent's context via parameters in the
mcp__agentic-payments__authorize_paymenttool (merchant, description, metadata) and themcp__agentic-payments__create_active_mandatetool (merchant_restrictions). - Boundary markers: The skill body does not contain instructions to use delimiters or ignore potentially malicious instructions embedded in the processed transaction data.
- Capability inventory: The agent can perform high-impact actions including cryptographic mandate signing (
mcp__agentic-payments__sign_mandate) and payment authorization (mcp__agentic-payments__authorize_payment). - Sanitization: There are no requirements or instructions for sanitizing, escaping, or validating the input data before it is interpolated into payment authorization requests.
Audit Metadata