financial-architect

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill defines a professional methodology for financial modeling. It includes explicit scope boundaries that protect the user by instructing the agent to refuse requests for accounting theory, tax advice, or investment recommendations.
  • [COMMAND_EXECUTION]: The skill uses a deterministic workflow where all mathematical operations are delegated to an external spreadsheet engine via companion scripts (idfa_ops.py and recalc_bridge.py). This is a safe and auditable execution pattern that prevents the AI from 'hallucinating' financial results.
  • [DATA_EXFILTRATION]: No unauthorized network activity or data exfiltration patterns were found. Operations are restricted to the user's spreadsheet environment.
  • [PROMPT_INJECTION]: No evidence of prompt injection or instruction override attempts was detected. The instructions prioritize model integrity and deterministic results.
  • [REMOTE_CODE_EXECUTION]: The skill does not download or execute arbitrary remote code. Dependencies like LibreOffice are well-known and used for local calculation verification.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 01:29 PM
Security Audit — agent-trust-hub — financial-architect