home-office-deduction

Pass

Audited by Gen Agent Trust Hub on Apr 14, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill utilizes internal tools like transaction_search and spending_summary to access sensitive financial data (mortgage, rent, utilities). This access is a functional requirement for tax calculation and no data is sent to external or unauthorized domains.
  • [PROMPT_INJECTION]: The skill processes external data from transaction records which constitutes a surface for indirect prompt injection.
  • Ingestion points: Transaction descriptions retrieved from financial logs.
  • Boundary markers: No explicit delimiters are used to wrap or isolate untrusted transaction text.
  • Capability inventory: The agent is limited to searching, summarizing, and exporting data; it lacks high-risk capabilities like arbitrary command execution or external network requests.
  • Sanitization: There is no evidence of validation or sanitization of the transaction content before it is processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 14, 2026, 01:49 PM