Security audit

guaikei-douyin-keyword-search-to-video-list

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent Douyin public-data retrieval tool, but it can collect and locally retain large sets of social-media comments, author data, and video metadata without a per-run save opt-in or retention controls.

Install only if you are comfortable sending Douyin queries, links, and a GUAIKEI API token to guaikei.com and keeping retrieved public social-media datasets in local JSON log files. Use it for legitimate public-data analysis, avoid collecting more comments or profile data than needed, and delete logs when they are no longer required.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Description-Behavior Mismatch

Medium

Confidence: 92% confidence
Finding: The CLI persists the full fetched Douyin results to a local JSON file after execution, even though the skill is presented primarily as a retrieval/analysis tool. This creates an undisclosed data-at-rest copy that may contain scraped profile/content metadata and can be exposed to other local users, backups, or later misuse without the operator realizing it.

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The README defines very broad natural-language triggers such as “搜一下”, “找”, and “热点”, plus fuzzy intent-priority rules that can match ordinary user phrasing. In an agent setting, this increases the chance of unintended tool invocation and silent collection of external data, especially because the tool can fetch and persist search, author, comment, and hotlist data.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The skill promotes automatic JSON export and later notes files are saved under a local logs directory, but the early description does not clearly warn that queried public video, author, and comment data is persisted to disk by default. This creates a data-handling transparency issue: users may trigger collection assuming ephemeral processing, while the system retains potentially sensitive research, monitoring targets, and comment datasets locally.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The documentation explicitly enables bulk collection of Douyin posts, author works, comments, and search results, but provides no guidance on privacy, lawful use, rate limiting, consent, or platform terms. In a social-media scraping and analytics skill, that omission increases the risk of misuse for surveillance, profiling, or collection of user-generated content at scale.

Missing User Warnings

Medium

Confidence: 83% confidence
Finding: The CLI writes fetched comment data to a local JSON file automatically after execution, without any explicit consent prompt, opt-in flag, or warning in this file. Because comment data may include personal or sensitive content and the filename is predictable, this behavior can create unintended data retention and privacy exposure on shared systems or in automated environments.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The code silently writes fetched results to a local JSON file with no explicit notice or confirmation in this file. Silent persistence increases privacy and compliance risk because users may expect transient CLI output, while the tool actually creates a durable local record of retrieved data.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.