afa-payments
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input and external data files which influence the generated risk assessments and are passed to a local calculation script.
- Ingestion points: User-provided metrics (TC40, TC15, TC05) and internal logs (learnings.jsonl).
- Boundary markers: Relies on external system protocols (interaction-protocol.md) for context separation, which are not explicitly defined within the provided files.
- Capability inventory: The skill can execute a local Python script (scripts/vamp_check.py) and generate complex financial and operational advice.
- Sanitization: Arguments passed to the Python script are validated as floats by the argparse library, mitigating simple string-based injection attacks at the script level.
Audit Metadata