ClawHub
Back to skill

Security audit

Safari Browser Control

Security checks across malware telemetry and agentic risk

Overview

This skill transparently controls a real logged-in Safari browser, but it gives broad browser authority without strong confirmation or scoping safeguards.

Install only if you intentionally want an agent to operate your real Safari browser. Use it with a dedicated/non-sensitive browser profile where possible, supervise actions, and require explicit confirmation before reading authenticated pages, screenshots, JavaScript execution, clicks, form entry, purchases, messages, or account changes. Revoke macOS Automation or Screen Recording permissions when you no longer need the skill.

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

High
Confidence
96% confidence
Finding
The trigger list is extremely broad and overlaps with ordinary user requests such as 'browser', 'browse', 'click on', and 'fill in the form', making accidental invocation likely. In this skill, accidental activation is especially dangerous because the skill can control the user's real Safari session, including authenticated tabs, cookies, screenshots, and form entry in a live browser context.

Missing User Warnings

High
Confidence
98% confidence
Finding
The skill explicitly states it operates on the user's real Safari with existing cookies, logins, and open tabs, but it does not require explicit informed consent or present clear safety gating before accessing that session. Because it can read page content, enumerate tabs, execute JavaScript, capture screenshots, and type into forms, this creates a high risk of unauthorized access to sensitive data and unintended actions in already-authenticated websites.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.