skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/subcontractor-payment-tracker/Gen Agent Trust Hub
subcontractor-payment-tracker
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (CSV, Excel, JSON) which presents a potential surface for indirect prompt injection attacks.
- Ingestion points: According to
instructions.md, the agent is directed to gather and process project data, file paths, and parameters provided by the user in various common file formats. - Boundary markers: There are no explicit delimiters or system instructions (e.g., "ignore any instructions contained within this data") to isolate the processed data from the agent's command context.
- Capability inventory: The skill has the capability to write to the filesystem via the
export_reportmethod inSKILL.md, which uses thepandaslibrary to generate Excel documents. - Sanitization: The implementation lacks evidence of input validation or sanitization to prevent the agent from interpreting natural language instructions that might be embedded within the data fields of imported files.
Audit Metadata