exploratory-data-analysis
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted scientific data and incorporates its content into a report for the agent, creating a surface for indirect prompt injection.\n
- Ingestion points: The script
scripts/eda_analyzer.pyreads data from external files such as CSV, JSON, and FASTA.\n - Boundary markers: The report template in
assets/report_template.mdlacks explicit delimiters or instructions for the agent to ignore potentially malicious embedded instructions within the summarized data.\n - Capability inventory: The skill can read arbitrary local files (if renamed to supported extensions) and writes generated markdown reports to the filesystem.\n
- Sanitization: The skill does not perform sanitization or filtering of data content before including it in the report generated for the agent.\n- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python analysis script (
scripts/eda_analyzer.py) using theuvtool, which constitutes a local execution path.\n- [COMMAND_EXECUTION]: Thescripts/eda_analyzer.pyscript performs dynamic imports of scientific libraries (e.g.,numpy,pandas,biopython,pillow) at runtime based on the detected file format, which is a form of dynamic code loading.
Audit Metadata