ClawHub

Chrome Mcp

Security checks across malware telemetry and agentic risk

Overview

This skill openly controls a real logged-in Chrome browser, but its account-changing power is broad and lacks clear consent boundaries.

Review before installing. Use this only if you want an agent to control a real logged-in Chrome session. Prefer a separate Chrome profile with only the accounts needed, keep remote debugging disabled when not actively using it, and require explicit confirmation before posting, deleting, purchases, form submissions, account changes, messages, or JavaScript execution.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger examples are extremely broad everyday requests like "帮我看看 X 上有什么新消息" and "打开 xxx 网站帮我...", which can cause the skill to activate in routine contexts without making the user aware that it will control a real, already-authenticated browser session. In this skill, that broad activation surface is especially dangerous because the connected browser can read private data and perform state-changing actions on logged-in sites.

Missing User Warnings

High
Confidence
99% confidence
Finding
The description advertises control of a real logged-in Chrome session, including posting, deleting, and operating any authenticated website, but provides no warning about privacy, account integrity, or destructive actions. Because this skill can access existing sessions and execute arbitrary browser actions/JS, missing warnings materially increase the risk of unintended account takeover-like behavior, data exposure, or irreversible user actions.

VirusTotal

58/58 vendors flagged this skill as clean.

View on VirusTotal