validate-data
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFE
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to ingest and validate external artifacts like reports, SQL queries, and spreadsheets. This creates a surface where instructions embedded in processed data could potentially influence agent behavior. The skill includes methodology checks and validation stances that encourage treating findings as evidence rather than instructions, which helps mitigate accidental obedience to data-embedded prompts.
- Data Access Patterns: The workflow involves inspecting source artifacts via paths, links, or dashboards. While this involves accessing external data sources, the skill context emphasizes using standard analytical tools (e.g., Jupyter notebooks, SQL connectors) to perform these checks. This is consistent with the intended purpose of verifying data integrity and calculation accuracy.
- Boundary and Sanitization Guidance: The skill provides a structured output format for validation reports, including sections for methodology, issues, and caveats. This structure acts as a natural boundary by separating the analysis of the data from the data itself, aiding in safe processing practices.
Audit Metadata