skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/as-built-documentation/Gen Agent Trust Hub
as-built-documentation
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill facilitates legitimate construction document management. The Python logic provided in the documentation uses established libraries such as pandas and openpyxl for data organization and reporting. The requested filesystem permissions are appropriate for the task of exporting project registers.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied project data from CSV, Excel, or JSON formats. Ingestion points: User-provided data paths or parameters in instructions.md. Boundary markers: Delimiters are absent, but instructions specify structured processing. Capability inventory: Filesystem write permission (export_register in SKILL.md). Sanitization: Present, as instructions.md explicitly requires validating inputs before processing. The focus on structured data management significantly mitigates the risk of prompt injection from data content.
Audit Metadata