Security audit

kuai-shou-competitor-research

Security checks across malware telemetry and agentic risk

Overview

This skill coherently fetches public Kuaishou data through a disclosed third-party API and saves results locally, with no evidence of hidden or destructive behavior.

Install only if you are comfortable sending Kuaishou keywords, profile or video URLs, request limits, and your GUAIKEI_API_TOKEN to guaikei.com. Treat collected comments and creator data as platform/user data, keep the generated logs directory protected, and use the tool only for public data you are authorized to analyze.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (41)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 99% confidence
Finding: 代码片段仅包含 scripts/utils/args.js，一个通用命令行参数解析器。它处理 --help、flag/alias、默认值、必填参数、布尔/数字校验、位置参数与帮助文本生成。这属于基础工具层实现，而不是快手公开数据检索或竞品分析功能本身。根据给定代码，既没有网络请求、也没有快手相关 API/页面访问、也没有结构化数据抽取或分析逻辑。因此该代码块的实际行为与声明描述存在明显不匹配。

Tp4

High

Category: MCP Tool Poisoning
Confidence: 97% confidence
Finding: The declared purpose is focused on Kuaishou public data collection and analysis. The actual code shown does not interact with Kuaishou, perform searches, fetch videos/posts/comments, or produce structured analysis data. Instead, it validates inputs, sanitizes filenames, creates directories, and writes content to local files under a logs folder. While logging can be a supporting detail in a larger system, this code chunk by itself implements a materially different capability—filesystem logging—rather than the declared Kuaishou retrieval functionality.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 99% confidence
Finding: The declared purpose describes a network/data-collection skill focused on Kuaishou public content and competitor analysis. The actual code chunk is a simple utility that accesses the local filesystem to read package.json and expose the package name. This is materially different from the declared primary purpose and capabilities. While utility files can be supportive, this chunk shows none of the claimed behavior and instead accesses a different resource class (local package metadata).

Ae1

High

Category: analysis-evasion
Content: | 查某个关键词的快手视频 | `scripts/kuaishou/search-cli.js` | `keyword` | 视频列表、作者信息、互动信息、跳转链接 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 查某个关键词的快手视频 | `scripts/kuaishou/search-cli.js` | `keyword` | 视频列表、作者信息、互动信息、跳转链接 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 看某个快手博主最近发布了什么 | `scripts/kuaishou/post-cli.js` | 博主主页 URL / user_id | 博主公开作品列表 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 看某个快手博主最近发布了什么 | `scripts/kuaishou/post-cli.js` | 博主主页 URL / user_id | 博主公开作品列表 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 看某个快手博主最近发布了什么 | `scripts/kuaishou/post-cli.js` | 博主主页 URL / user_id | 博主公开作品列表 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 看某篇快手视频的评论数据 | `scripts/kuaishou/comment-cli.js` | 视频 URL | 该视频的评论内容、评论者信息、互动数据 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 看某篇快手视频的评论数据 | `scripts/kuaishou/comment-cli.js` | 视频 URL | 该视频的评论内容、评论者信息、互动数据 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: | 看某篇快手视频的评论数据 | `scripts/kuaishou/comment-cli.js` | 视频 URL | 该视频的评论内容、评论者信息、互动数据 |
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: **👉 详细选项说明**, 可参阅 [完整选项说明](references/options.md)
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: **👉 详细选项说明**, 可参阅 [完整选项说明](references/options.md)
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: - 更新记录：见 `references/changelog.md`
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Lp3

Medium

Category: MCP Least Privilege
Confidence: 70% confidence
Finding: Without declared permissions the skill's intent is opaque and cannot be validated.

Natural-Language Policy Violations

Medium

Confidence: 86% confidence
Finding: The natural-language instructions, usage guidance, and examples are predominantly presented in Chinese, which can impose a specific language on users or downstream agents. The file does include some English fragments, but it does not explicitly offer a user language choice or state that Chinese is optional.

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: This markdown file documents commands to fetch up to 10,000 comments from Kuaishou videos, which can affect privacy expectations and involve bulk collection of user-generated data. The description provides usage syntax and examples but includes no warning or disclosure about handling platform data responsibly or respecting privacy and applicable terms.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: This code sends the provided video URL and limit to external comment APIs via `createCommentTask` and `getCommentTask`, then later saves the returned results to a local JSON file. While there is status logging, there is no clear user-facing disclosure that input data will be transmitted to a remote service and that retrieved data will be persisted locally.

Description-Behavior Mismatch

Medium

Confidence: 84% confidence
Finding: The manifest frames the skill as searching Kuaishou public data and outputting structured data for analysis scenarios. In addition to returning results on stdout, this file derives a filename from the URL and persists the full output locally via `log.taskWrite`, which is extra behavior not stated in the manifest description.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The script persists the full command output to disk via `log.taskWrite`, creating a file named from the target URL/user identifier. Although the skill logs progress messages, there is no explicit disclosure that data will be written to a local file, which is a safety-relevant side effect for a CLI that handles fetched account data.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The script persists the full search output, including the requested keyword and returned results, via log.taskWrite(). Although the script shows general status messages, it does not explicitly disclose that data will be saved to disk, which is a user-impacting file write operation.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The manifest describes a skill for Kuaishou public data retrieval, comment fetching, and competitor analysis, which is primarily a data collection/analysis function. This helper implements persistent local file writing to a logs directory, which is not part of that stated end-user purpose and introduces a broader capability than the manifest suggests.

Natural-Language Policy Violations

Medium

Confidence: 94% confidence
Finding: Multiple error messages are hard-coded in Chinese, including parse/network failures and token guidance, with no indication that the user can choose language or locale. This creates a natural-language locale policy issue because the skill appears to force one language for user-visible messaging.

Missing User Warnings

Medium

Confidence: 87% confidence
Finding: The code performs outbound HTTPS GET and POST requests and includes a TOKEN header, while POST also serializes and sends request data. In this file there is no confirmation prompt, user-facing log/print, or comment/docstring warning that user or system data may be transmitted to a remote service.

Natural-Language Policy Violations

Medium

Confidence: 98% confidence
Finding: This code emits all validation and error messages exclusively in Chinese, such as at L05, L10, L14, L18, L23, L50, L54, L59, and L64. The file provides no mechanism to select language or document a justified locale restriction, which creates a natural-language locale policy violation under the stated rule.

VirusTotal

60/60 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical

Code: suspicious.exposed_secret_literal
Location: SKILL.md:15