ClawHub
Back to skill

Security audit

K8s Toolkit Free

Security checks across malware telemetry and agentic risk

Overview

The skill is a Kubernetes troubleshooting guide, but it asks for broad agent tools and includes live-cluster commands without clear safety boundaries.

Review before installing. Use this only as a Kubernetes reference, and do not let an agent run kubectl, exec into pods, or scale deployments unless you explicitly choose the target cluster, namespace, and command. Prefer read-only diagnostics first and verify credentials are limited to the intended environment.

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 (5)

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The skill is presented as a troubleshooting knowledge base, but it declares both write and exec capabilities and states support for create/query/export operations. This mismatch can cause an agent or user to grant more privilege than necessary, increasing the chance of unintended file changes or command execution in a Kubernetes environment.

Intent-Code Divergence

High
Confidence
97% confidence
Finding
The file claims the free edition is a pure Markdown knowledge base requiring no cluster connection, yet multiple sections instruct running kubectl commands against live clusters. This contradiction can mislead users and agents into trusting the skill as offline-only while it actually encourages interaction with production infrastructure.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The activation language is overly broad, covering generic code generation, programming help, testing, and deployment scenarios beyond narrowly scoped Kubernetes troubleshooting. Overbroad triggering increases the chance the skill is invoked in unrelated contexts where its exec-capable guidance may be inappropriate or risky.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The capability coverage section enumerates an extremely broad set of keywords as valid activation context without meaningful constraints. In an agent environment, this can cause accidental activation and expose users to operational guidance or command execution paths when they did not explicitly request Kubernetes actions.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The markdown provides kubectl and deployment-management commands, including exec and scale operations, without clearly warning that they may impact live systems or expose sensitive runtime state. In cluster contexts, even seemingly routine commands can alter workloads, reveal secrets, or normalize unsafe operator behavior if presented without safeguards.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.