skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/cwicr-takeoff-helper/Gen Agent Trust Hub
cwicr-takeoff-helper
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-supplied data in CSV, Excel, and JSON formats, creating an attack surface for instructions embedded in data.
- Ingestion points: External project data files and direct user parameters described in instructions.md.
- Boundary markers: No explicit delimiters or "ignore" instructions are defined for the processed data in the agent instructions.
- Capability inventory: The skill has filesystem permissions enabled in claw.json and uses the pandas library to write Excel files in SKILL.md.
- Sanitization: Data is used in mathematical calculations, but there is no specific validation to prevent LLM subversion via data content.
Audit Metadata