israeli-car-accident-claim

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a bundled Python script, scripts/claim_router.py, to calculate insurance routing logic based on user input. This script is executed locally using the authorized Bash tool and contains no networking or high-risk operations.- [EXTERNAL_DOWNLOADS]: The skill provides links to official Israeli government websites (gov.il), legal databases (nevo.co.il), and non-profit legal guidance portals (kolzchut.org.il) to support its educational content.- [PROMPT_INJECTION]: The skill ingests user accident descriptions which are processed to extract parameters for claim routing. While this creates an indirect injection surface, the use of a local script with strict argument validation (argparse with choices) effectively mitigates risk.
  • Ingestion points: User-provided accident details.
  • Boundary markers: Not explicitly defined for user input processing.
  • Capability inventory: Local script execution via the Bash tool (scripts/claim_router.py).
  • Sanitization: Input is validated against specific allowed values within the routing script.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 03:58 PM
Security Audit — agent-trust-hub — israeli-car-accident-claim