ClawHub
Back to skill

Security audit

沟通三部曲

Security checks across malware telemetry and agentic risk

Overview

This skill is a simple communication helper with a disclosed trial/subscription check, and the reviewed files do not show hidden data collection or destructive behavior.

Install only if you are comfortable with the subscription model. If you add an API key, use one intended only for this provider and expect periodic verification with kaiwu8.com after the free trial expires. Remove the API_KEY value and the local .ttl file when uninstalling or resetting the license state.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Context-Inappropriate Capability

Medium
Confidence
91% confidence
Finding
The skill embeds a license/TTL enforcement mechanism that stores local state and is designed to contact a remote verification service. While this is not inherently malicious, it introduces behavior unrelated to the core communication function and creates a dependency on external authorization logic that can disable the skill or alter behavior outside the user's expectations.

Context-Inappropriate Capability

Medium
Confidence
95% confidence
Finding
The main communication entrypoint calls TTL verification before performing its actual task, which can trigger external network access unrelated to message analysis. This expands the attack surface, leaks usage metadata, and allows remote service availability or policy decisions to block core functionality.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The TTL verification code transmits an API key and skill identifier to a remote server without any user-facing notice or consent at the point of use. Sending credentials and usage-related metadata silently is a privacy and security concern, especially because the remote call is triggered by ordinary skill execution rather than an explicitly disclosed licensing action.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.