Security audit

guaikei-douyin-pull-video-comments

Security checks across malware telemetry and agentic risk

Overview

This Douyin data tool is not destructive, but it can trigger on broad generic requests and automatically stores scraped results locally, so users should review its scope before installing.

Install only if you want generic short-video research requests to be eligible for Douyin collection. Use clear prompts that name Douyin, avoid collecting or retaining sensitive comment/profile data unless authorized, and periodically delete the skill's logs directory when exported data is no longer needed.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (10)

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The CLI persists fetched comment data to a local JSON file after completing a retrieval operation. Comment content can contain personal data, sensitive opinions, or business intelligence, so silently writing it to disk increases data-retention and unintended disclosure risk, especially on shared systems or agent runtimes where workspace files may be accessible to other components.

Description-Behavior Mismatch

Medium

Confidence: 85% confidence
Finding: The script writes full search results and request metadata, including the user's keyword, to a local JSON log file without any visible consent or retention controls. This creates unnecessary data persistence that can expose potentially sensitive research terms, operational queries, and returned content to other local users, backups, or later compromise of the host.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The description says the skill should be used even when the word 'Douyin' is not present, which broadens invocation beyond explicit user intent. That can cause the agent to route generic research or content-analysis requests into a data-collection skill unexpectedly, increasing the chance of unintended external queries and over-collection of third-party data.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The implicit trigger examples ('竞品调研', '高赞内容', '最近什么热门') are generic requests that overlap with many normal research tasks. In an agent environment, this can lead to over-broad automatic invocation of the Douyin collection tooling, causing unintended data access, unnecessary token use, and user-surprising behavior.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The skill includes broad natural-language trigger examples such as generic phrases for searching or viewing content that could match ordinary user requests even when the user did not explicitly ask to use this Douyin scraping capability. In an agent setting, this can cause over-invocation of a data-collection skill and unintended collection of third-party content or comments, increasing privacy and compliance risk.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The intent-recognition rules define very broad priority keywords like '热搜 / 热点 / 榜单 / 今天什么火' and '搜索 / 搜一下 / 找 + 关键词' without requiring strong platform or task boundaries. This makes accidental triggering likely, especially because the skill metadata already says it may be used even if the word '抖音' is absent, so ordinary research requests could be routed into external scraping behavior.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The README advertises automatic JSON export and later states that comment, video, and account-derived data are saved into local log files, but it does not present a prominent privacy warning, retention policy, or handling guidance for potentially personal data contained in comments and profile-related metadata. This can lead to silent persistence of scraped data beyond the user's immediate task, creating privacy, data-minimization, and accidental disclosure risks.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The program writes full comment results to a local JSON file without an explicit user-facing warning or consent in this CLI path. Because this is a data-retrieval skill, users may reasonably expect transient output only; silent persistence creates privacy, compliance, and data-handling risks if comments contain personal or sensitive content.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The CLI persists fetched Douyin results to a local JSON file automatically, but there is no explicit user consent, warning, retention policy, or control over where sensitive scraped data is stored. In this skill context, the data may include creator/profile-derived content and operational research outputs; silent persistence increases the risk of unintended disclosure through shared workstations, backups, artifact collection, or later exfiltration from disk.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: Persisting query data and returned search content locally without explicit warning increases privacy and data-handling risk beyond the user's apparent expectation of a retrieval-only command. In the context of a Douyin intelligence/operations tool, search terms may reveal business intent, monitoring targets, or sensitive investigations, making silent persistence more dangerous.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.