ClawHub
Back to skill

Security audit

VSCode节点工具(免费版)

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed remote VSCode/Cursor control helper, but it can write or delete files through connected IDE nodes and its activation scope is broader than the level of control it provides.

Install only if you intend agents to operate a connected VSCode/Cursor workspace through the nodes gateway. Keep gateway allowCommands narrow, prefer read-only commands by default, and require explicit user approval before file writes, edits, deletes, renames, or any operation against sensitive repositories.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
79% confidence
Finding
The skill advertises an `exec` tool alongside broad remote IDE and node invocation workflows, which expands capability beyond narrowly scoped file/language/Git actions into general command execution. In an agent setting, this can enable arbitrary local or remote commands, increasing the risk of filesystem damage, credential exposure, or host compromise if the agent is prompted unsafely.

Intent-Code Divergence

Low
Confidence
84% confidence
Finding
The documentation states that absolute paths and `../` traversal are blocked, but SKILL.md itself does not implement or enforce those restrictions. This creates a dangerous trust gap: users or downstream agents may rely on protections that do not actually exist, enabling unintended access to files outside the workspace if the backing tool or wrapper fails open.

Vague Triggers

Medium
Confidence
76% confidence
Finding
The trigger condition is broad enough to match routine development requests, making the skill likely to activate in many contexts without clear user intent for remote IDE control. Because the skill includes write/delete and command-invocation capabilities, overbroad activation increases the chance of unintended destructive or privacy-impacting actions.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill describes file write, edit, delete, and remote command-style capabilities without a prominent warning about potential data loss, repository changes, or execution impact. In an agent workflow, omission of these warnings can cause operators to treat the skill as low-risk and approve actions that modify code, delete files, or affect connected environments.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.