Security audit

天猫电商视频生成与编辑

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed AI Hive video-generation helper that uploads user-selected media and stores an API key for that service, with some dormant generic client code that is not exposed by the normal skill command.

Install only if you are comfortable giving this skill an AI Hive API key and sending selected product images, videos, audio, and prompts to AI Hive/its upload storage. Prefer using a scoped API key, review ~/.ai-hive/config.json permissions, and choose an output directory deliberately when downloading generated media.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (6)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 90% confidence
Finding: The skill invokes a Python script with capabilities for shell execution, filesystem access, environment access, and network activity, but declares no permissions. This creates a transparency and least-privilege failure: users and reviewers cannot accurately assess what the skill may access or do, increasing the risk of unexpected data access, command execution, or exfiltration if the underlying script is unsafe or compromised.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 94% confidence
Finding: The documented purpose is narrowly about Tmall video generation, but the detected behavior includes broad generic capabilities such as chat, image generation, model enumeration, wallet/user-info queries, media upload, and API-key/browser setup. That mismatch is dangerous because it conceals materially broader access and functionality than users would expect, which can enable data collection, credential handling, unauthorized uploads, or account probing under the cover of a benign media-editing skill.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The file is presented as a Tmall video-generation/editing skill, but it embeds a much broader generic AI Hive client that supports text chat, image generation, model listing, uploads, and account inspection. This unnecessary capability expansion violates least privilege and increases the attack surface, because any agent or workflow invoking the skill can access functions unrelated to the advertised purpose.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill exposes a user-info endpoint even though its stated purpose is Tmall video generation/editing. Account metadata and wallet/balance information are sensitive operational data, and exposing retrieval of that data through an unrelated skill creates unnecessary information disclosure risk if the skill is misused or invoked by a broader agent.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill exposes a generic text-chat capability that is unrelated to the advertised Tmall video-generation/editing function. This enables the skill to act as a general-purpose LLM proxy, broadening the data-handling scope and making it easier for prompts or agent chains to send arbitrary text and images to the backend outside the intended business context.

Context-Inappropriate Capability

Low

Confidence: 87% confidence
Finding: The ability to enumerate models is broader than necessary for a fixed-purpose Tmall video skill. While lower severity than direct data access, model discovery can reveal backend inventory, available capabilities, and operational details that aid misuse and conflict with the principle of minimizing exposed functionality.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.