news-analytics-trading

Pass

Audited by Gen Agent Trust Hub on Apr 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script scripts/news_analytics_trading_diagnostics.py as part of its core workflow to process quantitative data.
  • [DATA_EXFILTRATION]: The diagnostics script reads from local CSV files and writes summary results to JSON files. While focused on performance metrics, this creates a handling surface for local file data.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via processed data files.
  • Ingestion points: The script scripts/news_analytics_trading_diagnostics.py reads untrusted data from external CSV files using pd.read_csv.
  • Boundary markers: No boundary markers or instructions are used to distinguish data from instructions within the processed CSV content.
  • Capability inventory: The skill possesses file read and write capabilities through the pandas library and standard file I/O operations.
  • Sanitization: Input data, including column headers and values, are processed and returned in a summary JSON without sanitization or escaping, allowing potentially malicious strings to enter the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 25, 2026, 06:11 AM
Security Audit — agent-trust-hub — news-analytics-trading