ClawHub
Back to skill

Security audit

LLM Deploy

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed documentation-only skill for deploying vLLM on GPU servers over SSH, with operational risks users should manage before running commands.

Before using this skill, verify the target SSH host, account, model path, port, and any llm-deploy script source. Run checks before deployment, avoid production servers until tested, monitor the tmux service, and stop it when no longer needed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The README promotes one-command deployment of a GPU-hosted model service and emphasizes speed and simplicity, but it does not warn users that deployment can start network-accessible services, consume substantial GPU/CPU/memory resources, or conflict with existing ports and workloads. In an agent/skill context, documentation strongly shapes autonomous or semi-autonomous execution behavior, so omission of operational and security warnings increases the chance of unsafe deployment on production or shared infrastructure.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal