Security audit

短剧预告片原创复刻｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE workflow for planning and generating short-drama trailer remakes, with user-controlled API use, media uploads, and local video edits.

Install only if you intend to use AI-HIVE for short-drama marketing production. Review prompts and pricing before submitting generation jobs, use only media you are authorized to upload, and avoid storing the API key in the config file if you prefer environment-only credentials.

SkillSpector

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (4)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 91% confidence
Finding: The skill documents executable commands and capabilities that access environment variables, local files, the network, and shell tooling, but it declares no permissions. This creates a hidden-capability problem: hosts or reviewers may underestimate what the skill can do, and users may be induced to run scripts that can handle secrets, upload local media, or modify files without an explicit permission model.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 89% confidence
Finding: The documented behavior extends beyond the stated purpose of trailer-remake planning into generic account inspection, model enumeration, generic chat/media generation, local credential setup, and arbitrary ffmpeg-style editing. This mismatch is dangerous because it broadens the operational surface and can mislead users into granting trust or running commands they would not expect from the advertised workflow, including actions involving credentials, billing-sensitive API calls, and local file manipulation.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: Implicit invocation allows the skill to be auto-selected without the user explicitly choosing it, which can trigger specialized behavior in contexts the user did not intend. In this skill, that increases the chance of unsolicited transformation of requests into AI-HIVE workflows and code, including for sensitive marketing or remixing scenarios where authorization and policy boundaries matter.

Natural-Language Policy Violations

Medium

Confidence: 82% confidence
Finding: The default prompt hard-codes Chinese output and a specific operational framing without checking the user's language preference. While not directly dangerous on its own, it can reduce transparency and informed consent, making it easier for the skill to steer users into a preset workflow they may not fully understand.

Static analysis

No suspicious patterns detected.