audit-payment-system
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze untrusted source code within a project environment, creating a potential surface for indirect prompt injection attacks where malicious code snippets could attempt to influence the agent's logic.\n
- Ingestion points: The skill uses
rg(ripgrep) to search through local codebase files as defined inSKILL.mdandreferences/checklist.md.\n - Boundary markers: While it instructs the agent to quote specific lines, it lacks specific instructions for delimiting or strictly isolating untrusted code content during processing.\n
- Capability inventory: The skill possesses read-only capabilities via
rgand documentation tools likeStripe MCP. It explicitly mandates a read-only posture and prohibits the creation of exploits or payment-fraud proofs-of-concept.\n - Sanitization: The instructions do not define specific sanitization or filtering logic for the data ingested from the source code search results.
Audit Metadata