Security audit

AI大模型专家｜短剧视频

Security checks across malware telemetry and agentic risk

Overview

This skill is purpose-aligned for AI-HIVE video generation, with disclosed media upload and local API-key use, but users should understand those external data and cost implications before installing.

Install only if you are comfortable sending prompts and selected media files to AI-HIVE and using an AI-HIVE API key for potentially billable video generation. Avoid uploading sensitive or unlicensed media, prefer environment-variable keys if you do not want persistent storage, and revoke or rotate the key if the local config file is exposed.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 88% confidence
Finding: The skill advertises and documents capabilities including environment-variable access, local file read/write, network access, and shell execution, but does not declare permissions or provide a least-privilege boundary. This creates a transparency and consent problem: users may invoke a skill that can access credentials, local files, and the network without explicit permission disclosure, increasing the risk of unintended data exposure or abuse if the underlying script is modified or misused.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 95% confidence
Finding: The documented purpose is narrowly framed as short-drama video generation, but the finding indicates broader behavior such as chat, image generation, model enumeration, account/wallet retrieval, credential setup, and general media upload. This mismatch is dangerous because users may trust the skill with media and credentials under a narrow use case while the implementation exposes additional privileged functions that can collect account data, broaden exfiltration paths, or perform actions outside informed user expectations.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill explicitly states that it will automatically upload user-provided media to AI-HIVE, but the description lacks a prominent privacy and data-handling warning. This is dangerous because users may submit sensitive images, videos, or audio without clear notice that the content will be transferred to a third-party service, potentially retained, logged, or used under platform policies they did not review.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The initialization flow writes API credentials to a local config file, but the instructions do not prominently warn users about local credential storage or its security implications. Storing long-lived API keys on disk can expose them to other local processes, backups, misconfigured permissions, or accidental sharing, especially if users do not understand where the key is stored and how to revoke or rotate it.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The skill enables implicit invocation with no trigger constraints, exclusions, or user-confirmation guardrails. In this context, the skill can automatically initiate AI-HIVE workflows involving media upload, task submission, polling, and result download, which increases the chance of unintended external actions, data transfer, or cost-incurring operations from ambiguous user requests.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.