signoz-explaining-alerts

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes alert metadata, including names, descriptions, and labels, along with query configurations fetched from the SigNoz platform. This ingested data is external and potentially untrusted. A malicious user with the ability to modify alerts in SigNoz could insert instructions designed to influence the agent's output when the skill identifies and explains those alerts.
  • Ingestion points: Metadata is retrieved through the signoz_get_alert, signoz_list_alert_rules, and signoz_get_alert_history tools.
  • Boundary markers: No specific boundary markers or instructions to ignore embedded commands are used when the agent processes alert configuration details.
  • Capability inventory: The skill is restricted to read-only SigNoz tools for fetching alert details and history.
  • Sanitization: The instructions do not mandate sanitizing or escaping the text content of alerts before it is presented in the final explanation.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 06:37 AM