skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/verification-loop-construction/Gen Agent Trust Hub
verification-loop-construction
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external construction data from BIM/CAD files, CSV, Excel, and JSON formats, creating a potential surface for indirect prompt injection.
- Ingestion points: Input data, file paths, and project parameters provided by users or read from the filesystem as defined in instructions.md.
- Boundary markers: The instructions do not define specific delimiters for untrusted content, though they recommend input validation.
- Capability inventory: The skill has filesystem access granted in claw.json and performs automated data verification and reporting tasks.
- Sanitization: The agent is explicitly instructed to validate inputs before processing according to instructions.md.
Audit Metadata