sdd-init

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill intentionally scans user home and project-level skill directories (including ~/.claude, ~/.gemini, ~/.copilot, etc.) and READS referenced files, then MANDATORILY persists the collected context to project files (.atl/skill-registry.md) and/or an "engram" backend without any mention of redaction or consent — posing a high risk of unauthorized data collection/exfiltration of sensitive files, though I found no obfuscated code, remote shells, or explicit remote-network exfiltration instructions.