database-enumeration

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs trying, recording, and returning plaintext credentials (examples and commands with -p'pass', default creds like SCOTT/TIGER, and "Return credentials for reuse testing"), requiring the LLM to handle and output secret values verbatim.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This skill contains explicit instructions for remote code execution (xp_cmdshell, UDFs, COPY PROGRAM, Redis config/write), backdoor installation (writing webshells, injecting SSH authorized_keys), and credential/data harvesting (dumping DB users, collecting creds), which are high-risk malicious behaviors if used without authorization.

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

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs actions that modify the target system state and achieve remote command execution or persistent access (enable xp_cmdshell, UDF/COPY PROGRAM, write webshells and authorized_keys via Redis), which are privileged/state-changing and thus compromise the machine.