guaikei-douyin-public-videos-dataset

• Outcome: Review
• Latest audit: Aug 21, 2026 · 1:45 AM UTC
• Version: 1.0.0

Overview

This skill collects public Douyin data as advertised, but broad automatic activation and default saved logs create review-worthy privacy and control concerns.

Install only if you are comfortable with agent-selected Douyin lookups, third-party API use through guaikei.com, and automatic local JSON logs of returned public data. Prefer confirming ambiguous requests before use, keep GUAIKEI_API_TOKEN private, and periodically clean the logs directory if results may contain personal or sensitive comments.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (10)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 82% confidence
• Finding: The skill explicitly depends on an environment variable (`GUAIKEI_API_TOKEN`) but does not declare corresponding permissions/capabilities in a machine-readable way. This creates a transparency and review gap: an agent may access sensitive environment data without users or platform policy clearly understanding that requirement.

Description-Behavior Mismatch

Medium

• Confidence: 86% confidence
• Finding: The CLI silently persists the fetched Douyin post results to a local JSON file after printing them, creating an undisclosed secondary data store. In a scraping/search skill, this increases data retention and disclosure risk because collected content may include metadata or other information the user did not expect to be saved on disk, especially on shared systems or agent runtimes.

Description-Behavior Mismatch

Medium

• Confidence: 91% confidence
• Finding: The CLI persists full search output to a local JSON file, which expands a read/search operation into local data retention. That can expose user search terms and returned content to other local users, backups, or later processes without an explicit need or disclosure, especially if results may contain sensitive or proprietary research activity.

Vague Triggers

High

• Confidence: 95% confidence
• Finding: The instruction to trigger the skill even when the user did not mention Douyin broadens activation beyond clear user intent. That can cause unsolicited transmission of user queries to this external data service, leading to privacy leakage, tool mis-selection, and unexpected use of third-party APIs.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The natural-language trigger mappings use broad phrases such as general search verbs and common topical requests, which can cause the skill to activate in conversations that are not actually requesting Douyin data collection. In an agent setting, this can lead to unintended third-party data access, unnecessary token-backed API calls, and user surprise because the README explicitly encourages invocation even when the user did not mention Douyin.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The intent-priority rules map generic words like '热点', '热搜', and '榜单' directly to the hot-list command, even though those terms are common across many domains and platforms. This increases the chance of overbroad activation and unintended use of an external data-collection capability, especially because the skill description says it should be used even when the user does not mention Douyin.

Missing User Warnings

Medium

• Confidence: 80% confidence
• Finding: The README promotes automatic JSON log export and local retention of collected video, author, comment, and hot-list data, but does not provide a clear privacy warning, retention policy, or handling guidance for potentially sensitive personal data in comments and creator metadata. In this skill context, large-scale collection plus local persistence makes privacy and data-governance risks more significant than a typical transient search tool.

Missing User Warnings

Low

• Confidence: 84% confidence
• Finding: The CLI automatically writes fetched comment data to a local JSON file without an explicit opt-in or warning. Comment datasets can contain personal data, usernames, and sensitive opinion content; silent persistence increases the risk of unintended retention, leakage to other local users/processes, or accidental inclusion in backups and source control.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The code writes full result data to a local JSON file without any user-facing warning or consent mechanism in this flow. This can expose scraped content and request context to other local users, backups, or downstream processes, and it broadens the data footprint beyond the expected CLI output.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The program writes search results to a local JSON log without any user-facing consent or warning in the execution flow. Silent persistence can leak user interests, research topics, and returned dataset contents beyond the immediate session, which is especially relevant for a browse/search skill users may expect to be ephemeral.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.