margin-management

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the system date command to generate formatted timestamps for recording entries in the margin dashboard.
  • [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection as it processes untrusted data from external balance files.
  • Ingestion points: Financial data is ingested from notebooks/updates/Balances_for_Account_{account_id}.csv and fin-guru/data/user-profile.yaml.
  • Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the data processing flow.
  • Capability inventory: The skill has the capability to write to external spreadsheets using the mcp__gdrive__sheets tool.
  • Sanitization: The skill extracts specific numeric fields but lacks explicit sanitization or validation of the input strings before processing or writing to the spreadsheet.
  • [SAFE]: The skill implements a manual safety checkpoint that halts the workflow and requests user confirmation if a margin draw exceeds $5,000, providing protection against unintended large financial transactions.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 09:49 AM