Security audit

Batch Files

Security checks across malware telemetry and agentic risk

Overview

This is mostly a Windows batch scripting reference, but it includes under-scoped Windows login/recovery bypass and persistence guidance that users should review carefully before installing.

Install only if you need an advanced Windows batch/CMD reference and are comfortable reviewing sensitive Windows administration material. Do not use the recovery-account, OOBE bypass, AutoRun/Run-key, hidden-console, credentialed FTP, or identity logging examples unless you are authorized to administer the affected systems and have tested the commands in a safe environment.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Context-Inappropriate Capability

High

Confidence: 96% confidence
Finding: The reference includes explicit account-access abuse guidance in the Recovery Environment section, such as replacing utilman.exe with cmd.exe and creating a new administrator account from the login screen. That is a well-known privilege-bypass/persistence technique unrelated to ordinary batch authoring and materially increases the chance the skill could facilitate unauthorized access.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The OOBE section gives step-by-step instructions to bypass Windows setup requirements and local-account restrictions. While not inherently code-execution malware, this is unrelated to a batch-file skill's core purpose and can enable policy circumvention or deployment of unmanaged systems.

Vague Triggers

Medium

Confidence: 83% confidence
Finding: The skill advertises very broad trigger phrases such as general automation, shell scripting, and scheduled task requests, which can cause it to activate in many loosely related contexts. For a high-capability Windows batch skill, overbroad routing increases the chance the agent will select it for sensitive operational tasks, expanding exposure to destructive commands, persistence mechanisms, or unsafe script generation when a narrower skill would be more appropriate.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The file is an extremely broad capability catalog covering persistence, registry edits, recovery manipulation, networking, account operations, scheduled tasks, and more, without clear activation boundaries or exclusions. In an agent skill, such breadth increases the risk that the model will surface hazardous or out-of-scope instructions for benign user requests.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: This section documents destructive or security-sensitive operations—deleting old files, registry changes, startup behavior, OOBE bypass, and recovery actions—without consistent warnings, confirmation requirements, or recovery guidance. Users or an agent may apply these snippets directly and cause system damage, data loss, or policy violations.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The file describes AutoRun, Run keys, startup locations, AppCompat registry changes, and related persistence/configuration mechanisms as ordinary techniques, with little warning about persistence abuse or system impact. In an agent context, that normalization can lead to covert startup persistence or unsafe configuration drift.

Ssd 3

Medium

Confidence: 83% confidence
Finding: The login-script guidance recommends persistent logging of computer names, usernames, IP addresses, and MAC addresses. That creates a privacy and security risk because it encourages collection of identifying network telemetry without discussing minimization, retention, or access controls.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.