skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/cwicr-data-validator/Gen Agent Trust Hub
cwicr-data-validator
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 untrusted external data sources such as Excel, CSV, and JSON files, which represents an attack surface for indirect prompt injection.
- Ingestion points: Data enters the agent's context through the
validate_dataframeandvalidate_estimatemethods inSKILL.mdwhich iterate over user-provided records. - Boundary markers: The skill implementation lacks specific delimiters or guardrail instructions (such as "ignore embedded instructions") to prevent the agent from being influenced by malicious content within the data records.
- Capability inventory: The skill possesses filesystem write capabilities via the
export_validation_reportmethod inSKILL.md, which writes data back to an Excel file. - Sanitization: While the code performs numeric range and data type validation, it does not sanitize text-based fields for potential prompt injection vectors or control characters.
Audit Metadata