kpi-dashboard

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly reads the external file ../../CLAUDE.md "on first use" to obtain scene-level configuration (KPI thresholds/alerts) that will directly control the agent's prompts/behavior at runtime, so ../../CLAUDE.md is a required runtime dependency.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)