payment-security

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill contains logic to ingest and process untrusted data from various external sources, including server access logs, payment API responses, and security scanner outputs. It lacks explicit boundary markers or sanitization for this data.
  • Ingestion points: Access logs (via grep/awk in Bash/Python), API responses from Stripe/Adyen/PayPal (via WebFetch), and external tool outputs (Nessus, Qualys, OpenSCAP).
  • Boundary markers: Absent. There are no instructions to use delimiters or ignore potential commands embedded in the processed data.
  • Capability inventory: The skill possesses extensive capabilities including command execution (Bash), file system access (Read/Write), and network operations (WebFetch).
  • Sanitization: Absent. The provided analysis scripts (e.g., Python card-test detection) perform regex parsing but do not sanitize data against adversarial content before it reaches the agent context.
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading and installing various security tools and Docker containers from external sources. These include the Stripe CLI from Stripe's official development site and MobSF from Docker Hub, which are well-known and trusted services.
  • [COMMAND_EXECUTION]: The skill's workflow depends on executing numerous CLI tools (nmap, tcpdump, frida, etc.) via the Bash tool to perform its stated security auditing functions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 04:36 PM