RootCauseAnalysis

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill contains a forced, quiet network callback to a suspicious localhost port (31337) executed before any other action plus configurable local-loading behavior and persistent execution logging — a pattern that can be abused as a backdoor or covert telemetry/exfiltration trigger.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). [Source: ~/.claude/LIFEOS/USER/CUSTOMIZATIONS/SKILLS/RootCauseAnalysis/; runtime path: skill reads user-provided customization files via “load and apply” before workflow execution, which exposes outsider-authored free text to the LLM.]