ClawHub
Back to skill

Security audit

MiniMax H3 电商带货视频 | LinkPix

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly aligned with paid AI video generation, but it normalizes unsafe API-key handling and broad host setup steps.

Review before installing. Only use this skill if you are comfortable sending selected product media and prompts to qhkit/Qinghu services and consuming paid credits after confirmation. Do not paste API keys into chat; configure the token through a local environment variable or platform secret mechanism instead. Prefer preinstalled, pinned tooling over allowing the skill to install or upgrade global packages at runtime.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill tells the agent to install Node.js, globally install qhkit, and perform upgrades on the host. That expands the skill from invoking an existing video tool into modifying the execution environment, which increases supply-chain and persistence risk if a package, mirror, or install path is compromised. In this context the behavior is framed as setup convenience, but it still exceeds least-privilege expectations for a media-generation skill.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The skill directs the agent to transform local files and, if needed, install Pillow or sharp-cli to do so. This adds extra software execution and file-processing behavior unrelated to the core declared capability, increasing attack surface through package installation and unintended handling of local user content. Because the files are user-provided images, the context lowers severity somewhat, but the environment modification remains risky.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The activation text contains broad product-video and advertising keywords and says the skill 'must trigger' for a wide set of requests. Overbroad triggering can cause the agent to invoke this skill in situations where the user did not intend to use this vendor/tooling, which can lead to unnecessary external API use, data disclosure to third parties, or unexpected billing. In a skill that can submit chargeable generation jobs, loose scoping is materially riskier.

Ssd 3

High
Confidence
98% confidence
Finding
The skill explicitly instructs the agent to ask the user to send their API key in chat and then use it in a command. Collecting secrets through natural-language chat creates a high risk of credential exposure via conversation logs, model retention, transcript sharing, or accidental reuse in later prompts. This is especially dangerous because the key grants access to a third-party paid API and the flow normalizes unsafe secret-handling behavior.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.