escalating-windows-privileges

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs dumping and decrypting stored credentials (LSASS/mimikatz, SAM/System dumps, GPP cpassword with gpp-decrypt, runas/savecred usage), which requires extracting and supplying secret values verbatim to tools/commands, creating a direct exfiltration risk.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). These are reputable, well-known GitHub repositories and the official LOLBAS reference (not typosquats or redirectors), so they are not inherently malicious sources, but they explicitly distribute exploitation tools, scripts and examples that can be used to build or deliver malware and thus represent a significant misuse risk if downloaded and executed without proper authorization and safeguards.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This content provides explicit, actionable steps and tooling to perform unauthorized privilege escalation, credential theft, persistent backdoors, and remote command execution (e.g., msfvenom/netcat reverse shells, registry autoruns, scheduled tasks, mimikatz/LSASS/SAM dumping, token impersonation and UAC bypass techniques), and is clearly intended to enable system compromise and remote access.

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 the agent to perform invasive privilege-escalation techniques (service binary/modification, registry edits, scheduled tasks, DLL hijacking, replacing system binaries, disabling defenses, credential dumping, UAC bypasses) that modify system files, change service configurations, and obtain elevated permissions, so it directly pushes compromising the host state.