business-investment-advisor
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data via JSON input files across its primary toolset.
- Ingestion points: Local JSON data files containing investment opportunities, portfolio holdings, and checklist scores are read by
scripts/investment_screener.py,scripts/portfolio_analyzer.py, andscripts/due_diligence_checklist.py. - Boundary markers: Absent; the output generated by the tools is structured text that the agent consumes without explicit delimiters or instructions to ignore embedded content within the data values.
- Capability inventory: The tools are strictly limited to data processing, mathematical scoring, and printing results to standard output; they do not possess network access, file system modification capabilities, or subprocess execution.
- Sanitization: The scripts do not perform sanitization or escaping of string values retrieved from the input JSON before presenting them in the agent's context.
Audit Metadata