ClawHub

AHTV PK to Xunlei

Security checks across malware telemetry and agentic risk

Overview

This skill is a narrow, disclosed workflow for finding specific AHTV episodes and adding them to a user's Xunlei Cloud account, with expected but sensitive login and account-mutation steps.

Install only if you are comfortable letting the agent run the included Python resolver scripts and operate a logged-in Xunlei Cloud browser session. Verify that login happens on the real Xunlei domain, share SMS codes only in that login flow, and review the final list of files added, skipped, moved, or renamed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding
The skill description promises cloud-drive login, deduplication, and file-management behavior that are not actually implemented in the referenced code path, creating a significant trust gap between what operators expect and what the skill can truly do. Security review and deployment decisions often rely on declared behavior; when implementation diverges, hidden or missing actions can bypass normal scrutiny, leading to unsafe execution patterns or operational misuse.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs collection and use of a phone number and SMS code for login without any warning, consent language, minimization guidance, or handling restrictions for these sensitive authentication factors. In an agent context, this increases the risk of credential interception, over-collection, accidental retention in logs, or phishing-like user interaction around one-time codes.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal