pyrolite

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied CSV data through the geochemistry_plots.py script and various geochemical functions. This creates a surface for indirect prompt injection. Ingestion points: Data is read via pandas.read_csv in the load_data function of scripts/geochemistry_plots.py. Boundary markers: No explicit delimiters or boundary instructions are used to distinguish data from instructions for the agent. Capability inventory: The skill's capabilities are limited to data calculations, visualization with matplotlib, and saving plot files. No network or shell execution capabilities are exposed. Sanitization: No validation or sanitization is performed on input data column names or values before processing. \n- [DYNAMIC_EXECUTION]: The provided script scripts/geochemistry_plots.py utilizes dynamic loading of the pyrolite library. Evidence: The script performs an import pyrolite inside the main() function after data loading. This is a common pattern for this library to enable pandas accessors but qualifies as dynamic loading of executable content at runtime.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 11:57 AM
Security Audit — agent-trust-hub — pyrolite