loki-mode

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). Several URLs point to direct GitHub release downloads and raw files from a single, non-enterprise GitHub user (asklokesh) and the repository's install instructions explicitly curl/unzip/run code (including an autonomous runner), which is a common vector for distributing malicious binaries or scripts if the source isn't verified.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill deliberately encourages running fully autonomous agents with permission-bypassing flags, disables or weakens audit/enterprise controls, allows broad filesystem access by default, and auto-loads cross-project learnings — collectively creating clear opportunities for stealthy autonomous actions and sensitive data exfiltration across projects.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The prompt explicitly instructs launching with "claude --dangerously-skip-permissions" (bypass permissions) and enforces "ZERO human intervention" / "NEVER wait for confirmation", which together encourage circumventing security controls and performing privileged or destructive actions autonomously.