Security audit

Shopify 电商视频生成与编辑

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI Hive video-generation helper that uses user-provided media and credentials for its stated Shopify video purpose, though it carries extra dormant generic client code.

Install only if you are comfortable giving this skill an AI Hive API key and uploading selected product images, videos, or audio to AI Hive. Prefer using the documented generate, task, upload, and init commands, and review charges before rendering variants. The package would benefit from removing unused generic client code and adding an explicit permissions statement, but the inspected artifacts do not show hidden execution, destructive actions, credential exfiltration, or unrelated automatic behavior.

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 (7)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 93% confidence
Finding: The skill advertises executable workflows that use shell, filesystem, environment variables, and network access, but it does not declare permissions. That creates a transparency and containment problem: a user or host may approve a seemingly simple content skill while the underlying implementation can read local inputs, write outputs, and call external services. In this context, the commands invoke a Python script and remote video endpoints, so undeclared capabilities materially expand attack surface even if the stated use case is legitimate.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 89% confidence
Finding: The documented purpose is narrowly scoped to Shopify/DTC video generation and editing, but the analyzed behavior reportedly includes broader AI Hive operations such as text chat, image generation, model enumeration, account/wallet queries, and generic media upload. That mismatch is dangerous because users may grant trust and data access for a limited ecommerce-video workflow while the implementation can interact with unrelated services and sensitive account metadata. The storefront-video context does not justify account info retrieval or broad model/service access, which increases suspicion and potential for misuse.

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The file is implemented as a broad AI Hive client supporting text chat, image generation, model enumeration, media upload, and task polling, while the declared skill is specifically for Shopify video generation/editing. This scope mismatch increases attack surface and can let downstream agents invoke unrelated capabilities that access account metadata or send arbitrary prompts/media off-platform, violating least privilege for the advertised skill.

Context-Inappropriate Capability

Medium

Confidence: 91% confidence
Finding: The skill includes a user-info endpoint that retrieves account and wallet information unrelated to Shopify video generation. In an agent context, this broadens available actions beyond user expectation and may expose sensitive billing or account metadata if the skill is invoked by another workflow.

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: General text chat allows arbitrary prompt submission to the provider and is outside the stated Shopify video-editing purpose. In practice this can be abused as an unsandboxed general AI proxy, bypassing intended skill boundaries and causing unreviewed data egress to the remote API.

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: Standalone image generation is not necessary for a Shopify video-generation/editing skill as described, and it expands the tool into a broader media-generation surface. That increases the chance of misuse, unintended billing, and transmission of unrelated user inputs or assets to the provider.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The module docstring explicitly describes a general-purpose AI capability tool, which contradicts the narrower Shopify video-specific manifest. In an agent ecosystem, such mismatches are dangerous because policy engines and users may trust the manifest while the code exposes broader functionality.

Static analysis

No suspicious patterns detected.