claude-cli-agent

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 running automated scripts with a --dangerously-skip-permissions flag to avoid UI approval, which advocates bypassing security/permission mechanisms and therefore risks compromising the host state.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)