ClawHub
Back to skill

Security audit

K8s清单生成入门

Security checks for vulnerabilities and agentic risk

Overview

This Kubernetes manifest skill is mostly understandable, but it asks for read and command execution access while its instructions broaden into vague file, API, network, and generic development behavior beyond the stated purpose.

Review this skill before installing. It does not show malicious actions, but only use it when you explicitly want Kubernetes YAML generation or validation, and do not allow it to run arbitrary commands, contact external services, or handle real secrets unless you have checked the exact command and data flow.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The skill is presented as a Kubernetes manifest generator/validator, but the later '核心功能' section broadens it into generic file handling, API integration, information retrieval, and command execution. This scope expansion can cause an agent to invoke the skill for unrelated tasks and to grant it far more operational authority than users would reasonably expect, increasing the chance of unsafe command or data-handling behavior.

Intent-Code Divergence

Medium
Confidence
88% confidence
Finding
The file claims '纯本地生成与校验' and '免费版无需API Key', yet nearby sections discuss external API integration and HTTPS network communication. These contradictory trust signals can mislead operators into assuming the skill is offline-only, when the documentation leaves room for networked behavior and external data handling.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The description uses broad activation language such as code generation, programming assistance, debugging, testing, and deployment, which overlaps with many normal developer requests. In agent environments, such vague routing criteria can trigger the skill outside its safe Kubernetes-manifest niche, exposing exec-enabled behavior in contexts where the user did not intend it.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The capability coverage section lists vague and expansive keywords, including generic development and workflow terms, as if they are supported invocation triggers. This increases the chance of over-broad skill selection and accidental exposure of read/exec permissions for tasks unrelated to Kubernetes YAML generation.

Static analysis

No suspicious patterns detected.