detecting-email-account-compromise

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external audit log data, which presents an indirect prompt injection surface.
  • Ingestion points: Log data is ingested from local JSON files via the CLI in scripts/agent.py.
  • Boundary markers: No explicit markers are used to isolate untrusted data from processing logic.
  • Capability inventory: The script is limited to local file system operations (read and write) and lacks network or subprocess execution capabilities.
  • Sanitization: Input is parsed as structured JSON but does not undergo content-based sanitization for security contexts.
  • [SAFE]: The skill includes references to trusted libraries and documentation from Microsoft and the AzureAD GitHub organization, which are documented neutrally.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 02:48 AM