israeli-attendance-wage-checker

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes user-provided attendance data which is then passed to a Python-based calculation tool. Although the logic is limited to arithmetic, processing external data remains a known vector for indirect instructions.
  • Ingestion points: Shift data such as clock-in/clock-out spans are ingested from user input to compute overtime and rest-day premiums.
  • Boundary markers: The skill contains explicit legal disclaimers and scope instructions in both English and Hebrew to delineate the boundaries of the AI's role.
  • Capability inventory: The skill is configured with access to the Bash(python3:*) tool specifically to execute the bundled scripts/reconcile_hours.py script.
  • Sanitization: The Python script includes validation for time formats (HH:MM) and uses error handling to reject malformed numeric or temporal input before processing calculations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 02:21 PM
Security Audit — agent-trust-hub — israeli-attendance-wage-checker