Security audit

短剧分镜故事板工作室｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE storyboard workflow that uses user-supplied media, API keys, and paid generation only for its stated creative purpose.

Install only if you are comfortable using AI-HIVE for storyboard media production. Treat uploaded reference images, videos, and audio as third-party data disclosure to AI-HIVE, confirm pricing and route before generation, and prefer environment variables or a tightly protected config file for the API key.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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 (7)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 87% confidence
Finding: The skill documents executable capabilities including environment-variable access, filesystem interaction, network calls, and shell/ffmpeg usage without any declared permission model or scoped constraints. In an agent ecosystem, this can cause the skill to be invoked with more effective power than users or reviewers expect, enabling secret exposure, unintended local file access, or outbound requests if supporting scripts are run.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 92% confidence
Finding: The skill is presented as a storyboard generator, but the described/available behaviors extend into generic AI chat, model enumeration, user-info retrieval, media upload, and broad ffmpeg processing. This mismatch is dangerous because operators may approve or invoke the skill for a narrow creative task while it can access unrelated remote endpoints and perform materially broader data handling than implied.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The file is implemented as a general-purpose AI Hive client exposing chat, video generation, model listing, user-info, upload, and task polling, while the declared skill is a storyboard/image-generation skill. Capability overreach increases attack surface and enables workflows unrelated to the approved purpose, which can bypass policy or review assumptions about what the skill can do.

Context-Inappropriate Capability

Low

Confidence: 81% confidence
Finding: The code can query user account information and wallet balance through get_user_info(), which is not necessary for generating storyboard images. Even if not directly exposed by the wrapper path, retaining account-inspection functionality in the skill code broadens accessible sensitive metadata and weakens least-privilege boundaries.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The generic text chat path allows arbitrary model interaction beyond the storyboard/image-generation purpose. In a narrowly scoped content-production skill, this creates an unnecessary policy bypass surface because users or integrators may leverage the included capability for unrelated or higher-risk tasks.

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: The generic video-generation capability exceeds the wrapper's image/storyboard role and enables broader media creation than the manifest suggests. This mismatch matters because video generation may carry different safety, cost, and abuse implications, and reviewers or users may not expect it from this skill.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The skill enables implicit invocation without any narrowing conditions, so it may auto-activate on broad storyboard or video-generation requests even when the user did not clearly intend to use this tool. Because the skill can drive AI-HIVE workflows and produce runnable commands, unintended activation increases the chance of surprise external actions, unwanted cost-incurring operations, or generation of risky marketing/media content without explicit user confirmation.

Static analysis

No suspicious patterns detected.