ClawHub

Playwright Scraper

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed stealth web scraper, but it needs Review because it can bypass bot protections for any URL and passes user input into a shell command.

Install only if you understand the legal, account, and host-security risks of stealth scraping and will use it only on targets where you have permission. The publisher should remove shell interpolation, validate URLs, add authorized-use guidance, and include the missing runtime script before this is treated as routine.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Context-Inappropriate Capability

Medium
Confidence
97% confidence
Finding
The tool invokes a shell command via execSync using attacker-controlled input derived from the URL parameter. Even though the feature is framed as scraping, exposing subprocess execution creates a command-injection path if the URL contains shell metacharacters or quoting tricks, allowing arbitrary local command execution under the server's privileges.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill explicitly advertises anti-bot evasion and stealth scraping capabilities without any warning, restriction, or compliance guidance. That makes misuse easier for unauthorized scraping, circumvention of site protections, and violations of target-site terms or policy, especially because the sole tool is described as able to scrape any URL with bot-detection bypass.

Missing User Warnings

Medium
Confidence
99% confidence
Finding
A user-supplied URL is inserted directly into a shell command and executed synchronously, with no validation, escaping guarantees, or confirmation barrier. In this MCP skill context, the tool is remotely invokable by an agent, so a crafted input can lead to arbitrary command execution on the host rather than merely an unsafe scrape request.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal