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

• Outcome: Review
• Latest audit: Aug 21, 2026 · 2:34 AM UTC
• Version: 1.0.0

Overview

This skill mostly does what it claims, but it can activate on broad non-Douyin requests and automatically sends and stores research data without enough user control.

Install only if you are comfortable with an agent using this skill for ambiguous short-video or trend-research tasks, sending those queries and Douyin URLs to guaikei.com, and saving returned results locally. Prefer explicit user confirmation before use and periodically delete the logs directory if the research is sensitive.

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
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (9)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 84% confidence
• Finding: The skill requires access to an environment variable (`GUAIKEI_API_TOKEN`) but does not declare that capability in a permissions model. This creates a transparency and governance gap: an agent may invoke code that reads secrets from the environment without an explicit declaration, making review, sandboxing, and least-privilege enforcement harder.

Vague Triggers

High

• Confidence: 94% confidence
• Finding: The top-level description instructs the agent to use this skill even when the user did not explicitly mention Douyin, based on broad workflow cues like content research or topic planning. That can cause overbroad activation, unnecessary transmission or collection of third-party data, and user-surprising behavior in multi-step tasks where platform choice was never authorized.

Vague Triggers

High

• Confidence: 95% confidence
• Finding: The generalized trigger section explicitly broadens invocation to short-video research, competitor analysis, and topic ideation without sufficient constraints. In an agent setting, this can route unrelated or ambiguous requests into external data collection automatically, increasing privacy, compliance, and user-intent mismatch risks.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill states that logs are automatically saved to the local `logs/` directory, including collected result data, but does not provide an explicit user warning or consent mechanism for persistence. Persisting search terms, account URLs, comments, or trend data can create avoidable retention and privacy risk, especially in shared workspaces or agent environments.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The natural-language mapping includes very broad phrases such as "搜索 / 搜一下 / 找" that are common in ordinary conversation and are not tightly scoped to Douyin-specific tasks. In an agentic workflow, this can cause unintended invocation of the skill, sending user-provided queries to the external service and producing actions the user did not explicitly intend for this integration.

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: The intent-recognition priority rules use generic triggers like "热搜 / 热点 / 榜单 / 今天什么火" without constraining them to Douyin context. Because these phrases are broadly applicable to news, social media, or general trend queries, the agent may route unrelated requests into this skill and disclose user queries to the configured third-party API unnecessarily.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The README explains how to configure a third-party API token and run commands, but it does not clearly warn that user-supplied keywords, account URLs, video URLs, and comment retrieval requests are transmitted to an external service. This weakens informed consent and increases the risk of unintentional data sharing, especially when the skill is embedded in multi-step workflows and may activate implicitly.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The CLI writes fetched comment results to a local JSON file derived from the target URL without clearly disclosing that persistence to the user. Comment data can contain personal data, usernames, opinions, or other sensitive content, so silently storing it increases privacy, retention, and accidental exposure risk on shared systems or in downstream backups/log collection. In this skill context, the feature is more dangerous because it is explicitly used for bulk competitor research, comment analysis, and topic mining, which increases the likelihood of collecting large volumes of user-generated content.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The CLI writes the full search output, including user-supplied keywords and returned results, to a local JSON file without any user-facing notice or consent in this file. In the context of competitor research, comment analysis, and topic planning, those queries and results may reveal business strategy, account targets, or other sensitive operational data to other local users, backup systems, or unintended log collectors.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.