usgs-produced-waters

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches geochemical data from the official USGS ScienceBase repository (sciencebase.gov). These downloads are necessary for the skill's primary function and target a well-known, trusted government service.
  • [PROMPT_INJECTION]: The skill ingests large external datasets (CSV and Excel), which represents an indirect prompt injection surface where instructions potentially embedded in data records could influence agent behavior.
  • Ingestion points: ScienceBase CSV and Excel files downloaded to the /tmp/ directory as specified in SKILL.md and python_example.py.
  • Boundary markers: Absent; there are no explicit instructions to the agent to disregard instructions that might be contained within the data fields.
  • Capability inventory: Network data retrieval, local file system writes (caching), and narrative analysis generation based on data content.
  • Sanitization: Data is processed into numeric and text formats using standard libraries without specific sanitization of fields before they are used in narrative summaries.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 06:30 PM