Antfu Vueuse Functions

• Outcome: Review
• Latest audit: Jun 18, 2026 at 9:01 PM UTC
• Version:
  1.0.0

Overview

This VueUse guide is not malware, but it broadly tells agents to automatically use sensitive browser capabilities without enough scoping or user-consent guidance.

Install only if you want an aggressive VueUse-first coding helper. Review generated code carefully when it suggests browser permissions, file access, clipboard, location, screen capture, Bluetooth, notifications, injected scripts, workers, or persistent storage, and require explicit user-facing consent for those features.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Description-Behavior Mismatch

Medium

• Confidence: 93% confidence
• Finding: The skill is presented as a general VueUse helper, but it authorizes automatic use of powerful browser and device APIs far beyond harmless reactive utilities. That creates a pathway for an agent to introduce privacy-sensitive or capability-expanding code such as script injection, local file access, Bluetooth, notifications, screen capture, and worker execution without clear user intent or safety gating.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: Marking sensitive capabilities as AUTO encourages an agent to select them whenever they seem applicable, even when the user did not ask for privileged browser interactions. In a code-generation context, this can silently normalize invasive patterns and lead to privacy, data-handling, or supply-chain risks being embedded into application code.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The activation rule applies the skill to essentially any Vue.js or Nuxt development request, causing the broad and partially sensitive guidance in this file to be injected into many unrelated tasks. Because the skill includes automatic recommendations for privileged browser features, over-broad triggering increases the chance of unnecessary or unsafe capability use.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The instruction to 'always check first whether a VueUse function can implement the requirement' and to prefer AUTO use lacks limiting conditions, exclusions, or safety examples. Without negative examples, an agent may over-apply VueUse wrappers around sensitive browser APIs simply because they exist, rather than because the user intentionally requested those capabilities.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill lists many browser APIs with privacy or data implications but provides no warnings, consent requirements, or user-notice language despite marking them AUTO. In practice, this omission can cause generated solutions to access files, devices, location, clipboard, notifications, or display capture without the agent surfacing the risks to the user.

VirusTotal

63/63 vendors flagged this skill as clean.