Back to skill

Security audit

AI短视频工厂

Security checks for vulnerabilities and agentic risk

Overview

This is a Markdown-only short-video workflow skill whose external media, TTS, subtitle, and platform guidance are disclosed and aligned with making videos.

Before installing, be aware that this skill is optimized for Chinese short-video production and may call on external media, TTS, AI-generation, and music services during use. Confirm platform, language, licensing, budget, and whether any batch output or publication steps should happen before letting it proceed.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list includes broad, everyday phrases such as '帮我做视频' and context like '用户提到做视频/生成视频', which can match many benign conversations and cause unintended activation. In an agent system, overbroad activation increases the chance this skill takes over requests outside its intended scope, potentially causing unwanted workflow execution or routing errors.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The markdown trigger section repeats generic triggers like '做视频' and 'AI视频' without enough constraints, making accidental invocation likely. Because this skill can initiate a multi-step content-production flow, misfires could lead to irrelevant outputs, user confusion, and interference with other more appropriate skills.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

Line L047 states '需英文字幕或双语字幕', which is a natural-language locale policy constraint that forces English or bilingual output for a platform scenario without any user opt-in or justification. This can violate organizational language-choice policies because it removes the user's ability to choose their preferred language strategy.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
71% confidence
Finding

The natural-language description and invocation guidance are entirely written for Chinese-language use, which may indicate a default language constraint without explicit opt-in. While Step 1 later mentions confirming “语言”, the top-level skill description does not clearly state that users may choose their preferred language before output generation.

Content

No source excerpt is available for this finding.