Video Multi-Platform Publish

• Outcome: Review
• Latest audit: Sep 12, 2026 · 1:28 AM UTC
• Version: 1.0.0

Overview

The skill appears to be a local video clipping demo, but it materially overstates real publishing, analytics, and automation capabilities.

Treat this as a local video transcoding and publishing simulation tool, not a real multi-platform publisher. Do not rely on it for actual uploads or analytics, and do not provide platform credentials unless a future version clearly documents authenticated APIs, per-platform consent, and data-sharing behavior.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Behavioral AST: exec() Call, eval() Call, Dynamic Import

Findings (16)

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 97% confidence
• Finding: The declared description overstates the implementation. The clipping/transcoding portion substantially matches the claim of automatic platform-format clipping. However, the core claim of multi-platform publishing is inaccurate: publish_video only prints messages and marks success as simulated, with no API calls, authentication, browser automation, or network interaction to actually publish content. The claimed intelligent optimization/recommendation features are also not implemented as intelligence; title handling is just length truncation, tags are fixed per platform, and publish times are static constants. Additionally, the clipping stage outputs files named '<inputstem>_<platform>.mp4', while the publishing stage looks for 'video_<platform>.mp4', so the supposed full workflow will fail to find the generated files. This is a material description-to-behavior mismatch.

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 92% confidence
• Finding: The skill documentation advertises shell-based execution patterns such as running Python scripts and installing dependencies, but the skill metadata does not declare any tool scope or allowed-tools. In an agent environment, this can cause users or orchestrators to invoke shell access without explicit authorization boundaries, increasing the chance of unintended command execution.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill promotes multi-platform publishing and analytics collection without clearly warning that content, metadata, and possibly account-linked information may be sent to third-party platforms. That omission can cause users to trigger external posting or data sharing without informed consent, especially in an automated agent context.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The usage examples show direct commands and API calls that appear to publish content immediately, but they do not warn that the actions may create public posts on external platforms. In an automation setting, example-driven execution can lead to accidental disclosure, reputation damage, or unintended publication of sensitive media.

Sudo/Root Execution

Medium

• Category: Privilege Escalation

Content

# Install ffmpeg
winget install ffmpeg  # Windows
brew install ffmpeg    # macOS
sudo apt-get install ffmpeg  # Linux
```

---

• Confidence: 70% confidence
• Finding: Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill advertises one-click publishing, cross-platform sync, and analytics tracking, but does not disclose that it may require platform credentials, transmit content to third parties, or perform external account actions. This is dangerous because users may invoke the skill without understanding the scope of account access, data sharing, and potentially irreversible publishing operations across multiple services.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger keywords are broad enough to match common user requests about publishing or video workflows, which can cause this skill to activate in contexts where the user did not explicitly intend cross-platform publishing actions. In a skill that can prepare content for external platforms and potentially interact with publishing workflows, overbroad activation increases the risk of unintended data handling or external-side effects.

Natural-Language Policy Violations

Medium

• Confidence: 93% confidence
• Finding: This code file contains user-facing natural-language descriptions and console output in Chinese, starting with the module docstring and continuing throughout the CLI experience. The skill does not offer any language or locale selection, which can violate a policy requiring user opt-in before forcing a specific language.

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

def check_ffmpeg():
    """检查 ffmpeg 是否安装"""
    try:
        subprocess.run(['ffmpeg', '-version'], capture_output=True, check=True)
        return True
    except (subprocess.CalledProcessError, FileNotFoundError):
        return False

• Confidence: 70% confidence
• Finding: subprocess module calls execute external commands. Without careful input validation, this enables command injection.

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

]
    
    try:
        result = subprocess.run(cmd, capture_output=True, text=True, check=True)
        info = {}
        for line in result.stdout.strip().split('\n'):
            if '=' in line:

• Confidence: 70% confidence
• Finding: subprocess module calls execute external commands. Without careful input validation, this enables command injection.

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

# 执行
    print(f"  🎬 剪辑 {platform_info['name']} 版本...")
    try:
        subprocess.run(cmd, check=True, capture_output=True)
        print(f"  ✅ {platform_info['name']} 版本完成")
        return True
    except subprocess.CalledProcessError as e:

• Confidence: 70% confidence
• Finding: subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Description-Behavior Mismatch

Medium

• Confidence: 98% confidence
• Finding: The manifest describes 'one-click publishing' to WeChat, Bilibili, Xiaohongshu, Douyin, YouTube, TikTok and more, implying actual publication to external platforms. However, the implementation explicitly labels publishing as simulated, only prints metadata and success messages, and performs no network/API interaction with any platform.

Intent-Code Divergence

Medium

• Confidence: 91% confidence
• Finding: The function is documented as '发布视频到平台 (模拟)', but the surrounding behavior still prints '发布成功' for each platform without any actual upload or API call. This creates intent-level confusion because the tool presents simulated actions as completed platform publications.

Sudo/Root Execution

Medium

• Category: Privilege Escalation

Content

print("\n请安装 ffmpeg:")
        print("Windows: winget install ffmpeg")
        print("macOS: brew install ffmpeg")
        print("Linux: sudo apt-get install ffmpeg")
        sys.exit(1)
    
    # 检查输入文件

• Confidence: 70% confidence
• Finding: Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Natural-Language Policy Violations

Low

• Confidence: 81% confidence
• Finding: The document presents core workflow instructions primarily in Chinese while also including some English headings and examples, but it does not state that the skill is intended only for Chinese-speaking users or offer language selection. Under the policy, forcing a specific language without opt-in can be a natural-language policy concern unless the locale limitation is clearly documented and justified.

Natural-Language Policy Violations

Low

• Confidence: 87% confidence
• Finding: Natural-language policy review applies to all file types, including manifests. The user-facing description is presented only in Chinese, while the file otherwise mixes English identifiers, and it does not indicate that the skill is region-specific or offer language selection.

Static analysis

No suspicious patterns detected.