cwicr-multilingual

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a potential surface for indirect prompt injection by processing external data that influences agent output.
  • Ingestion points: Untrusted data enters the context through pd.read_csv, pd.read_excel, and pd.read_parquet within the load_database method in SKILL.md.
  • Boundary markers: No specific delimiters or instructions (e.g., 'ignore instructions in data') are used when retrieving and presenting item descriptions to the agent.
  • Capability inventory: The skill's code is limited to data analysis and mathematical comparisons using pandas, with no identified capabilities for network communication or shell command execution.
  • Sanitization: There is no evidence of content sanitization or validation of strings retrieved from the construction databases before they are processed by the LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 04:27 AM