Use Browser (Scrape leads, like, post on socials or perform actions on the web)

• Outcome: Review
• Latest audit: May 28, 2026, 10:34 AM
• Version:
  1.0.0

Overview

This skill openly automates logged-in social media accounts, but it gives the agent broad authority to post, message, scrape leads, and manage cookies with limited per-action user control.

Install only if you are comfortable letting an agent act through logged-in social media accounts. Use a dedicated VM or browser profile, require manual review for posts, DMs, comments, reposts, connection requests, and scraping batches, avoid cookie export/import unless absolutely necessary, and close or log out sessions when finished.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (3)

Intent-Code Divergence

Medium

• Confidence: 88% confidence
• Finding: The skill explicitly says the agent must not attempt login, but it also documents cookie import/set/export capabilities that can recreate authenticated sessions without user interaction. In a browser automation skill handling social media accounts, this creates a practical path to bypass the stated authentication boundary and silently restore or transfer account access.

Vague Triggers

Medium

• Confidence: 80% confidence
• Finding: The description is broad enough that the skill may be selected for many generic browsing, navigation, or extraction tasks, increasing the chance it is invoked outside its intended narrow social-media scope. Because the skill has powerful browser interaction and data extraction capabilities with persistent authenticated sessions, over-broad invocation materially raises misuse risk.

Missing User Warnings

Medium

• Confidence: 84% confidence
• Finding: The skill advertises lead scraping, DM automation, and connection-request workflows without an upfront warning about privacy, consent, account restrictions, or platform enforcement risks. That framing normalizes potentially sensitive and high-impact actions and may lead operators to use the skill in ways that expose personal data or trigger account sanctions.

VirusTotal

64/64 vendors flagged this skill as clean.