ClawHub
Back to skill

Security audit

Docker Essentials Tool Free

Security checks across malware telemetry and agentic risk

Overview

This Docker helper is mostly coherent, but it gives an agent broad shell authority for Docker operations and includes destructive cleanup examples without clear confirmation safeguards.

Install only if you want an agent to run Docker CLI commands on your machine. Before destructive actions such as prune, rm -f, rmi, volume rm, or docker-compose down -v, require the agent to show the exact target resources and get explicit confirmation, especially around volumes and private registry credentials.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (3)

Intent-Code Divergence

Low
Confidence
91% confidence
Finding
The error-handling table instructs the agent to execute a ping-based network connectivity test when required parameters are missing, which is unrelated to the stated Docker-management purpose. This expands behavior beyond the user's requested intent and can cause unnecessary command execution or host/network probing in environments where the skill should remain scoped to Docker operations.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger conditions are overly broad, saying the skill should be used for coding assistance, debugging, testing, and deployment in general. Because the skill has exec capability and contains many operational commands, broad activation criteria increase the chance it is invoked for loosely related requests and performs Docker or shell actions the user did not specifically intend.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill includes destructive examples such as force-removing containers and pruning resources without clear warnings, confirmation requirements, or safeguards. In an exec-capable agent, normalizing these commands without caution can lead to accidental data loss, service disruption, or irreversible cleanup of images, containers, and volumes.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.