ClawHub
Back to skill

Security audit

Token Stats Pkg

Security checks across malware telemetry and agentic risk

Overview

This is a local token-usage viewer that reads assistant history files and can optionally create a command shortcut, with no evidence of network exfiltration, credential use, or hidden destructive behavior.

Install only if you are comfortable with a local utility reading assistant usage/history files from Hermes, Claude Code, CodeX, and OpenClaw. Run setup only if you want the global token-stats command; remove ~/.local/bin/token-stats if you no longer want that shortcut.

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 (1)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The skill description promises a simple interactive selector that prompts every run, but the documented behavior includes additional capabilities such as direct backend selection, watch mode, exports, setup that writes executables to ~/.local/bin, and shell configuration inspection. This mismatch weakens informed consent: users may invoke or install the skill expecting passive local viewing while it also performs filesystem modifications and broader operational behavior.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.