skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/toolbox-talk-generator/Gen Agent Trust Hub
toolbox-talk-generator
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-provided context like work activities and incident reports to generate safety scripts. The Python code in SKILL.md demonstrates direct interpolation of this data into LLM prompts.
- Ingestion points: Project data files (CSV, Excel, JSON) and direct input as described in instructions.md.
- Boundary markers: No specific delimiters or ignore instructions are implemented in the code templates to separate user data from instructions.
- Capability inventory: The skill requests filesystem permissions in claw.json for data processing and export.
- Sanitization: No input validation or sanitization logic is provided in the reference implementation to handle potentially malicious context strings.
Audit Metadata