skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/erp-data-extractor/Gen Agent Trust Hub
erp-data-extractor
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data in formats like CSV, Excel, and JSON. This creates a surface where malicious instructions could be embedded in data files to influence the agent. The skill includes mitigation instructions requiring input validation.
- Ingestion points: User-provided project data and files as specified in instructions.md.
- Boundary markers: Assistant instructions require following the API and methods defined in SKILL.md.
- Capability inventory: The skill has filesystem write access for exporting data through export_to_excel and export_to_json.
- Sanitization: The instructions.md file explicitly directs the assistant to validate inputs before processing.
- [SAFE]: The implementation uses standard data science libraries (pandas, openpyxl) and does not contain network calls, obfuscation, or persistence mechanisms.
Audit Metadata