Security audit

dou-yin-data-insight

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed, read-only Douyin public-data research tool, with noteworthy but purpose-aligned risks around third-party API use, automatic local result logs, and collection of public comments/profile metadata.

Install only if you trust guaikei.com with the API token and requested Douyin queries. Use it for explicit Douyin public-data research, avoid collecting more comments or profile data than needed, and review or delete the logs directory if results may contain personal or sensitive audience information.

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 Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (10)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 89% confidence
Finding: The skill requires access to the GUAIKEI_API_TOKEN environment variable, but no explicit permissions model is declared to make that sensitive capability visible and reviewable. This creates a transparency and governance gap: operators may invoke the skill without realizing it can access and transmit secrets to a third-party API, increasing the risk of unintended data exposure or misuse if the skill is modified or misrouted.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 82% confidence
Finding: The declared description presents the skill as a competitor/content research tool, but the documented behavior extends to collecting comments, hot-list data, creator post histories, and saving results locally to logs. This mismatch is dangerous because users and reviewers may authorize the skill under a narrower use case than its actual data collection and retention behavior, which can lead to overcollection, privacy/compliance issues, and unexpected local persistence of scraped third-party data.

Description-Behavior Mismatch

Medium

Confidence: 91% confidence
Finding: This CLI explicitly retrieves comment data for arbitrary Douyin posts, which expands the skill from high-level competitive/content analysis into direct comment harvesting. In the context of a research-oriented skill, collecting user-generated comments introduces privacy, compliance, and secondary-use risks because it enables accumulation of audience data beyond what the manifest clearly discloses.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The skill defines very broad natural-language triggers such as '搜索 / 搜一下 / 找 ...' and maps them directly to command execution. In an agent setting, these common phrases can cause unintended invocation during ordinary conversation, leading to unrequested data collection or external API use. Because this skill performs scraping/analytics against a third-party service, accidental activation is more sensitive than a purely local utility.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The competitor-monitoring examples use broad phrases like viewing a creator's works/homepage/account, which can overlap with normal discussion about social media accounts. This increases the chance that an agent invokes bulk collection against a target account without clear user intent, potentially causing privacy, compliance, or misuse concerns in a competitive-intelligence context.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The hot-list trigger definitions include highly generic phrases such as '今天什么火' and '热点/榜单', which are common in everyday conversation and not uniquely tied to Douyin. This overlap can cause the skill to activate unexpectedly and perform external requests when the user may only be asking a general question, creating unnecessary data access and action without informed consent.

Missing User Warnings

Medium

Confidence: 81% confidence
Finding: The documentation explicitly supports collecting creator posts and comment data from Douyin, which can include personal data, user-generated content, and behavioral information, but it provides no privacy notice, lawful-use guidance, retention limits, or terms-of-service considerations. In a competitor research skill, this omission increases the chance that operators will scrape and process personal data in ways that are non-compliant or ethically unsafe.

Missing User Warnings

Low

Confidence: 92% confidence
Finding: The GET request places the token and search keyword into request parameters, which commonly end up in URLs, server logs, proxies, browser history, and monitoring systems. Exposing authentication material and potentially sensitive search terms through query strings increases the chance of credential leakage and privacy exposure even if TLS is used.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The code persists fetched comment results to a local JSON file without any explicit user notice, consent flow, retention limit, or access control visible in this file. Because the stored data may contain user-generated content and identifiers, silent local persistence increases the risk of unintended disclosure, over-retention, and reuse of scraped data.

Missing User Warnings

Medium

Confidence: 77% confidence
Finding: The script silently writes fetched results to a local JSON file after completing the request, without any explicit user warning, consent, or opt-out. Because the output may contain collected competitor/content data and request metadata, this can create unintended local data retention and exposure on shared systems or in automated environments.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.