smart-money-profiler
Pass
Audited by Gen Agent Trust Hub on Jul 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill's
prompt_templateinSKILL.mdcreates a surface for indirect prompt injection by directly interpolating untrusted data. - Ingestion points: User-provided
task(query) andattachments(file data) are injected into the agent's core instructions. - Boundary markers: The template lacks robust delimiters or instructions to disregard commands within the input, using only basic labels like '任务与材料' (Tasks and Materials).
- Capability inventory: The skill has the ability to modify the local file system (writing to
profiles/seats.json) and call external APIs via thepandadata-apidependency. - Sanitization: No validation or escaping of the user-controlled input strings is performed before prompt interpolation.
Audit Metadata