ClawHub
Back to skill

Security audit

无缝循环短视频生成器｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed AI-HIVE video-generation helper that uses an API key and user-selected media, with no evidence of hidden destructive or unrelated data collection behavior.

Install only if you are comfortable giving this skill an AI-HIVE API key, uploading selected media to AI-HIVE, and potentially incurring generation costs after parameter confirmation. Avoid using private, copyrighted, or unauthorized reference material unless you intend it to leave the machine for processing.

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill embeds executable command examples and operational guidance that use environment variables, local file paths, shell commands, network access, and file I/O, yet no explicit permissions are declared. This creates a capability/consent mismatch: an agent or reviewer may treat the skill as low-privilege while it actually instructs actions that can access secrets, upload local content, and invoke external APIs with possible billing and data exfiltration consequences.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The script exposes broad capabilities beyond the declared seamless-loop video skill, including generic chat, image generation, model enumeration, user-info lookup, and arbitrary media upload. In an agent skill context, this expands the attack surface and enables capability drift: a caller expecting a narrowly scoped video tool can instead access unrelated API functions and account metadata.

Intent-Code Divergence

Medium
Confidence
87% confidence
Finding
Describing the file as a general AI invocation tool conflicts with the specialized skill identity and can mislead reviewers, policy enforcement, and orchestration layers about the true scope of exposed functionality. This kind of scope misrepresentation is dangerous because it hides excess capabilities that could be abused through the skill interface.

Natural-Language Policy Violations

Medium
Confidence
96% confidence
Finding
The default prompt hard-codes 'production-ready Chinese workflow', which can override or bias the user's preferred language without explicit consent. This is primarily a policy and UX integrity issue: it may cause unexpected output behavior, reduce accessibility for non-Chinese users, and create prompt-injection-like steering at invocation time.

Static analysis

No suspicious patterns detected.