Claw Browser

Security checks across malware telemetry and agentic risk

Overview

This skill appears to be a disclosed Playwright-based browser helper whose higher-risk powers are expected for web browsing.

Install this only if you want an agent to drive a real browser. Avoid using it on sensitive local, intranet, account, or credential pages unless you explicitly intend that access, and review any browser-binary installation step before running it.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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)

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger guidance says to use the skill when a user asks to 'open' or 'see' a website, which is overly broad and can cause the agent to browse arbitrary user-supplied URLs without sufficient constraint or safety checks. In a browser-capable skill, loose activation criteria increase the chance of visiting phishing, local-network, or otherwise sensitive targets, especially because the tool renders dynamic content in a real browser.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal