payment-advisor
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues were detected. The skill follows best practices by separating logic into a deterministic engine and providing a clear security policy ([SECURITY_RULES.md]) for the agent to follow during code generation. All external documentation URLs for payment providers are legitimate and direct users to official vendor resources.- [INDIRECT_PROMPT_INJECTION]: The skill manages the attack surface for indirect prompt injection by using a deterministic ranking engine to process untrusted user and project data. It explicitly prohibits the LLM from computing scores manually.
- Ingestion points: Untrusted data enters the agent context through the interactive wizard and from external project analysis tools.
- Boundary markers: The skill uses structured JSON blocks for data exchange.
- Capability inventory: The skill utilizes [Bash] for local script execution, along with [Read] and [Glob].
- Sanitization: Untrusted input is encapsulated in JSON for script-based processing, minimizing direct LLM exposure to instructions within the data.
Audit Metadata