tracking-health

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [SAFE]: Analysis of SKILL.md and all files in the references/ directory confirms that the skill is purely informational and instructional. It does not contain code intended for execution by the agent beyond standard natural language processing of instructions.
  • [DATA_EXFILTRATION]: The skill acts as a security manual for tracking. In references/url-pii-protection.md, it details how to prevent PII from being exfiltrated via GA4 payloads. It recommends architecture-level fixes and GTM-level overrides to sanitize URLs before data collection.
  • [INDIRECT_PROMPT_INJECTION]: In references/geo/tw.md, the skill addresses the ingestion of untrusted data from the LINE Messaging API. It identifies the risk of forged events and mandates signature verification using HMAC-SHA256 to validate incoming data.
  • [CREDENTIALS_UNSAFE]: In references/ga4-implementation.md, the skill provides examples using placeholders for sensitive values and provides explicit warnings against hardcoding real secrets in the codebase or frontend environments.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 12:03 AM
Security Audit — agent-trust-hub — tracking-health