ClawHub
Back to skill

Security audit

kubectl Operations

Security checks for vulnerabilities and agentic risk

Overview

The skill clearly describes Kubernetes administration tasks, including commands that can change, stop, or delete workloads.

Use appropriately restricted Kubernetes permissions. Before running examples, verify the intended cluster, namespace, resource, and action; deletion and scaling to zero can interrupt services.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Activation keywords include everyday terms without an explicit Kubernetes-context requirement.

Medium
Category
Trigger Abuse
Confidence
98% confidence
Finding

The activation rules include generic terms such as "update", "change", and "run command" without explicitly requiring Kubernetes context, potentially routing unrelated requests into a workflow with cluster-modification and pod-execution capabilities. The Kubernetes-specific title, description, and state-checking guidance reduce the risk, but the activation rules do not themselves establish authorization for those operations; actual harm would depend on downstream agent behavior and available permissions.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.