data-analysis-pro

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data from various file formats (CSV, Excel, Parquet, JSON, ZIP) via the io_router.py and preprocessing.py scripts. This creates a surface for indirect prompt injection where malicious instructions embedded in data cells could potentially influence the agent's behavior during summary generation or reporting.
  • Ingestion points: scripts/io_router.py (data loading), scripts/column_profile.py (column profiling), scripts/eda_stats.py (statistical analysis).
  • Boundary markers: The skill lacks explicit boundary markers or structured delimiters when passing raw or summarized data content from scripts back to the AI agent's context.
  • Capability inventory: The toolkit includes broad capabilities such as file system writes (writing JSON results and PNG charts via common.py), environment variable modification, and executing complex statistical modeling via sklearn and statsmodels.
  • Sanitization: There is no evidence of content sanitization or instruction-filtering for data values processed during the analysis phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 05:32 AM