zabbix-problem-review

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides operational guidelines for interacting with a Zabbix monitoring system through a read-only interface. It includes specific rules to ensure the agent distinguishes between 'no active problems' and 'NMS unreachable' to avoid misleading users during connectivity issues.
  • [PROMPT_INJECTION]: No evidence of prompt injection, instruction overrides, or DAN-style jailbreak attempts was found. The instructions are focused on operational accuracy.
  • [DATA_EXFILTRATION]: The skill correctly uses environment variables (ZABBIX_TOKEN) for managing secrets, adhering to secure development practices. There are no patterns suggesting unauthorized data exfiltration or hardcoded credentials.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns, suspicious package installations, or unauthorized script downloads were identified. The skill relies on pre-configured environment tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 09:47 AM
Security Audit — agent-trust-hub — zabbix-problem-review