stock-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests breaking news headlines from external Google News RSS feeds to provide real-time alerts and influence its stock analysis recommendations. This creates a surface where manipulated external content could theoretically be used to trigger specific risk flags (e.g., 'Taiwan conflict' or 'Banking crisis') and bias the agent's financial advice.
- Ingestion points:
check_breaking_newsinscripts/analyze_stock.pyfetches data fromhttps://news.google.com/rss/search. - Boundary markers: The skill does not use specific delimiters or instructions to prevent the agent from being influenced by embedded commands within these headlines.
- Capability inventory: The skill can perform complex stock analysis, manage persistent portfolios in a local JSON file, and make multiple network requests to financial services.
- Sanitization: Content is checked for keyword matches and displayed in the output caveats without extensive filtering.
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to Yahoo Finance, CNN, SEC EDGAR, and Google News RSS to gather financial data and news. These are all well-known technology and financial service providers.
Audit Metadata