onboarding

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.70). The prompt includes explicit non-functional directives—e.g., "You are in AUTONOMOUS MODE. Do NOT ask questions." and a SELF-EVOLUTION TELEMETRY block that instructs checking ~/.claude/projects and appending telemetry—which are behavioral/telemetry actions outside the stated onboarding/doc-generation purpose and could persist or exfiltrate metadata.