Security audit

short-video-content-replicator

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed short-video processing workflow, but users should understand it runs local Python helpers, downloads media, and writes output files.

Install this only if you intend to run a local short-video download/transcription pipeline and have the required companion skills or helper scripts installed. Use explicit URLs and output directories, verify rights/privacy for third-party videos, and review the missing/mismatched helper-script references if you need dependable operation.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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 (8)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 94% confidence
Finding: The skill invokes multiple Python CLI scripts that download remote content and write files, yet it declares no corresponding permissions beyond requiring a Python binary. This creates a mismatch between documented capabilities and effective behavior, reducing auditability and increasing the risk of unintended shell/file/network actions being triggered without explicit user or platform review.

Intent-Code Divergence

Medium

Confidence: 84% confidence
Finding: The file header claims a different skill name ('universal-voice-cloner') than the implemented short-video replication workflow. In a security-sensitive agent ecosystem, metadata inconsistency can mislead reviewers and operators about what code is supposed to do, weakening trust boundaries and enabling risky code to hide behind unrelated labels.

Description-Behavior Mismatch

Low

Confidence: 77% confidence
Finding: The workflow executes multiple local helper scripts and cross-skill components that are not clearly reflected in the described component mapping. Hidden or undocumented execution dependencies increase supply-chain and review risk because operators may approve one workflow while unintentionally granting execution to additional code paths.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The trigger phrase "一键端到端处理这个视频" is broad enough to overlap with normal conversational requests and may cause unintended skill activation. In an agent environment, overly generic activation language can route unrelated user inputs into a powerful workflow that downloads and processes external media, creating consent, privacy, and misuse risks.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The README mixes precise commands like "/short-video-content-replicator https://..." with ambiguous natural-language prompts such as "帮我复制这个抖音视频的内容," without stating what minimum conditions are required to trigger the skill. This ambiguity increases the chance of accidental activation of a multi-step workflow that fetches and transforms third-party content, potentially leading to unauthorized processing or policy bypass.

Natural-Language Policy Violations

Medium

Confidence: 80% confidence
Finding: The workflow description indicates Chinese-first processing and correction behavior without making language selection an explicit user choice. This can cause inaccurate transcription, unwanted text normalization, or silent transformation of non-Chinese content, which is a safety and reliability issue when users expect faithful transcription.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The trigger phrases are broad enough to match ordinary requests about converting video to text, extracting audio, or processing media, which can cause the skill to auto-run in situations where the user did not intend a six-step workflow with downloading and file generation. In this context, overbroad activation is risky because the skill performs external retrieval and multi-stage processing rather than a narrowly scoped, read-only transformation.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill description does not prominently warn that it will download third-party video content from external URLs and create several output artifacts across directories. Without clear disclosure, users may unknowingly trigger network access, storage consumption, and potential handling of copyrighted or sensitive media, making accidental misuse more likely.

Static analysis

No suspicious patterns detected.