ML Training

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only ML training skill whose deployment and remote-task risks are disclosed enough to treat as normal cautions rather than a Review issue.

Before installing, be prepared to approve each Nautilus task, dataset, compute budget, and deployment target yourself. Do not let fetched tasks trigger cloud spend, public endpoints, credential use, or persistent model hosting without an explicit confirmation step and appropriate access controls.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill explicitly includes deployment capabilities such as wrapping models with FastAPI/Flask and publishing inference endpoints, but it does not warn users that these actions can affect external systems or expose services publicly. In an agent setting, this omission increases the risk that an agent autonomously deploys or publishes a model endpoint without clear user confirmation, creating unintended exposure, operational changes, or data leakage.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal