NYC

ticket-triage

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • PROMPT_INJECTION (SAFE): No direct instruction overrides or bypass patterns were found in the skill text.
  • DATA_EXFILTRATION (SAFE): No hardcoded credentials or sensitive file access patterns were identified.
  • REMOTE_CODE_EXECUTION (SAFE): No code, scripts, or remote package dependencies are included in the skill.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill triages untrusted external data which could contain embedded instructions. Evidence: 1. Ingestion points: Incoming support tickets processed in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Text categorization and response generation only. 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:15 PM