funding-collateral-borrow

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data through the scripts/funding_collateral_borrow_diagnostics.py script.
  • Ingestion points: Data is ingested from input.csv using pandas.read_csv in scripts/funding_collateral_borrow_diagnostics.py.
  • Boundary markers: Not explicitly defined in the file format, but the script enforces a strict column schema.
  • Capability inventory: The script performs statistical summarization and writes results to a local JSON file specified by the user.
  • Sanitization: The script employs pd.to_numeric(errors="coerce") and pd.to_datetime(errors="coerce") on all input data before calculation, effectively sanitizing the input by converting non-conforming data to null values, which prevents data-driven instruction injection into the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 08:45 AM
Security Audit — agent-trust-hub — funding-collateral-borrow