Security audit

guaikei-douyin-lookup-tool

Security checks across malware telemetry and agentic risk

Overview

The skill mostly matches its Douyin research purpose, but it needs Review because it can trigger on broad research prompts and automatically sends and saves large datasets with limited user control.

Review before installing if you may use this on sensitive research targets. Treat searches, Douyin URLs, comment datasets, and the GUAIKEI token as data sent to the provider and also saved locally under logs; delete logs when no longer needed and avoid ambiguous prompts unless you want this Douyin tool to run.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (10)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The trigger guidance says the skill should activate even when users do not explicitly mention Douyin, including broad requests like general short-video research or 'help me research this track.' That can cause unintended invocation in contexts involving other platforms or broader research tasks, leading to unnecessary data collection, wrong-tool execution, and possible privacy/compliance issues if the agent fetches external data the user did not clearly request.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The skill states that outputs are automatically saved to a persistent `logs/` directory with timestamps and query parameters, but it does not prominently warn users that their queries and collected data will be stored locally. Persistent storage of search terms, URLs, comments, or research targets can expose sensitive business interests or personal data to later users, operators, or other processes on the system.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The natural-language trigger examples are broad enough that ordinary user phrases like '找 AI 相关视频' or '最火的' may activate this skill outside a clearly bounded Douyin-analysis request. In an agentic environment, this can cause unintended tool execution and collection of external data when the user only meant to discuss or brainstorm, increasing privacy, compliance, and action-without-clear-consent risks.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The intent recognition rules rely on generic keywords such as '搜索', '热点', '评论', and '账号/博主', which are common in everyday conversation and not sufficiently scoped to this specific tool. That makes accidental or overbroad invocation more likely, especially because the rules also define priority ordering that could force a tool action based on partial matches rather than clear user intent.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The README advertises automatic JSON export of scraped video, author, comment, and hotlist data to local logs, but it does not prominently warn that potentially sensitive public data will persist on disk by default. In practice, retained comment datasets and scraped profile/video metadata can create privacy, data-minimization, and secondary-exposure risks if stored longer than expected or accessed by other local users/processes.

Missing User Warnings

Low

Confidence: 82% confidence
Finding: The GET request includes the token and search parameters directly in the query string. Query parameters are more likely to be exposed through server logs, intermediary logs, browser/debug tooling, and monitoring systems, which can leak credentials or sensitive usage data even if the request is otherwise legitimate.

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: The CLI writes fetched comment results to a local JSON file by default, and those results may contain personal or sensitive user-generated content. Silent persistence increases the risk of unintended retention, later disclosure to other local users/processes, or accidental inclusion in backups and logs, especially because this skill is explicitly designed for bulk comment collection and analysis.

Missing User Warnings

Medium

Confidence: 72% confidence
Finding: The script transmits the provided URL and limit to a remote API to create and retrieve comment tasks, but this data flow is not surfaced to the user in a clear consent or privacy notice. In a data-collection skill, undisclosed network transmission can create privacy, compliance, and trust issues, particularly if users believe processing is local or if URLs embed identifiers tied to research targets.

Missing User Warnings

Medium

Confidence: 76% confidence
Finding: The CLI silently writes the full fetched results to a local file without explicit user consent or an in-file warning. Even if the source data is nominally public, the saved dataset may contain aggregated content, metadata, or analysis targets that users do not expect to be retained on disk, creating privacy, operational, and data-handling risks on shared systems.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The CLI writes full search output to a local file whose name includes the user-supplied keyword, but the user is not clearly warned that data will be persisted on disk. In this skill context, search results and keywords may reveal sensitive business research, campaign plans, or analyst interests, so silent local logging increases the risk of unintended data disclosure to other local users, backup systems, or telemetry collectors.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.