Security audit

Javascript Skills Tool Free

Security checks across malware telemetry and agentic risk

Overview

This JavaScript style-guide skill is not malicious, but its broad auto-activation and command-capable scope are too loosely bounded for a simple advisory guide.

Review this before installing. It appears to be a prompt-only JavaScript style guide, but only grant it command execution in workspaces where you are comfortable with the agent running linting or related commands, and prefer explicit invocation for style review rather than automatic use on any JavaScript task.

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 (6)

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The skill is presented as a JavaScript style guide, but these sections expand its authority into generic operational modes such as create, modify, delete, export, import, and save. That mismatch can mislead an agent into performing filesystem or workflow actions beyond passive code-style guidance, creating capability confusion and increasing the chance of unintended state-changing behavior.

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: Claiming delete/reset/import/export/save capabilities for a style-guide skill introduces unjustified action verbs that may be interpreted by an agent as permission to alter user data or workspace state. Even without executable code, this kind of prompt-level overclaim broadens the skill's effective authority and can be abused through ambiguous user requests.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The document contains contradictory safety boundaries: one section says the tool only provides suggestions and code snippets, while earlier sections imply modify/delete/save/export behavior. This inconsistency is dangerous because agents may follow the broader authority model, and users cannot reliably predict whether the skill is advisory or mutating.

Vague Triggers

High

Confidence: 97% confidence
Finding: The auto-activation rule triggers whenever a request mentions JavaScript or frontend coding, which is far broader than the skill's narrow purpose. Overbroad activation can cause prompt injection by collision, where this skill inserts itself into unrelated tasks and steers agent behavior unexpectedly.

Vague Triggers

High

Confidence: 96% confidence
Finding: The trigger conditions include broad categories like code generation, programming assistance, debugging, testing, and deployment, without clear boundaries. Such vague routing criteria can make the skill activate in many unrelated contexts, creating prompt-surface expansion and increasing the risk of instruction interference with more appropriate tools or policies.

Vague Triggers

High

Confidence: 98% confidence
Finding: The quick-start says merely mentioning JavaScript will auto-activate the tool, which is an extremely weak trigger. In an agent environment, this can lead to accidental invocation across a large fraction of developer conversations, giving the skill undue influence over unrelated tasks and increasing the chance of unsafe or conflicting behavior.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.