command-code

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The docs enable and encourage running an autonomous, unattended agent with a global "bypass permissions" flag that inherits auth/environment and can run arbitrary shell commands and fetch remote skills — a high-risk capability that can be abused for RCE, data exfiltration, credential theft, and supply-chain attacks if misused.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill explicitly instructs using the --yolo/--dangerously-skip-permissions flag which bypasses permission prompts and allows an autonomous agent to run arbitrary shell commands and file edits unsupervised, effectively encouraging bypassing security controls and modifying machine state.