workspace-health-dashboard

Security checks across malware telemetry and agentic risk

Overview

This is a read-only workspace health dashboard, but its dependency-security wording is stronger than what the code actually checks.

Install only if you want a lightweight, read-only workspace health summary. Do not treat its dependency status as a real vulnerability audit; run a dedicated tool such as npm audit or another lockfile-aware scanner before making security decisions.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The function is documented as a dependency vulnerability check, and the module/manifest also frame this dashboard as consolidating dependency security. In practice, the code merely parses package.json, counts dependencies, and explicitly notes that it does not perform a real audit, which is a direct contradiction of the stated check purpose.

Intent-Code Divergence

Medium
Confidence
88% confidence
Finding
The code runs a protected-skills status check and stores its result, but then unconditionally increments the dashboard 'passed' count with the comment 'Always pass if no error'. This contradicts the intent of a status-based health check because missing protected skills can produce a warning result that is still reported as passed in the summary.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal