guaikei快手博主作品评论采集

• Outcome: Review
• Latest audit: Sep 28, 2026 · 12:48 AM UTC
• Version: 1.0.0

Overview

This package is labeled as a Kuaishou data scraper, but the inspected artifact actually processes and uploads arbitrary video files for remote transcription and content generation.

Review carefully before installing. If you want a video transcription and content-generation tool, this package mostly describes that workflow, but it should be renamed and scoped honestly. Do not install it expecting Kuaishou post/comment scraping, and only provide videos or local files you are comfortable uploading to the provider's remote service.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (46)

Description-Behavior Mismatch

Critical

• Category: Not specified by scanner
• Confidence: 100% confidence

Finding

The manifest and body materially disagree about what the skill does: a supposed Kuaishou scraper is actually a video/audio transcription and content-generation tool. This kind of identity mismatch is especially dangerous because it can cause users or agents to provide unrelated data, credentials, or trust approvals to a skill whose real behavior includes remote upload and off-platform processing.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High

• Category: Not specified by scanner
• Confidence: 99% confidence

Finding

The README is materially inconsistent with the declared skill purpose: it documents a generic video transcription and LLM-powered content processing product rather than a Kuaishou public-data scraping skill. This mismatch is dangerous because agents or users may invoke the skill under false assumptions, causing unintended data handling, overbroad processing, or execution of capabilities outside the approved scope.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The documented behavior significantly exceeds the advertised manifest by describing generic video upload, transcription, OCR, rewriting, translation, and prompt-driven analysis features. Scope expansion like this is risky because it can bypass trust boundaries and policy expectations for a narrowly scoped Kuaishou data skill, potentially enabling broader data exfiltration or unreviewed remote processing.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 100% confidence

Finding

This is a clear case of functional misrepresentation: the skill actually processes arbitrary videos, uploads content to a third party, stores local task state, and returns generated text, not Kuaishou comments/posts/search JSON. In an agent ecosystem, such deceptive packaging is dangerous because it can misroute sensitive inputs and bypass user expectations about data handling.

Content

No source excerpt is available for this finding.

Vague Triggers

High

• Category: Not specified by scanner
• Confidence: 97% confidence

Finding

The trigger description is overly broad and mismatched, making the skill eligible for many generic video-analysis requests rather than a narrow Kuaishou-data use case. Overbroad triggers are dangerous in agent environments because they increase the chance this misleading, higher-privilege skill is auto-selected for unrelated tasks.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The inline text repeatedly claims Kuaishou creator-post/comment collection while the actual command examples are for video-to-text and prompt rewriting. Repeated contradictory claims increase the chance of accidental invocation and hide the true data flow to a third-party processor.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High

• Category: Not specified by scanner
• Confidence: 99% confidence

Finding

The package metadata describes a video-to-text/transcription skill, while the declared skill purpose is Kuaishou public-data retrieval. This kind of identity mismatch is dangerous because users, reviewers, or orchestrators may invoke or trust the skill for one purpose while it actually implements another, enabling deceptive capability smuggling or accidental execution of unrelated code paths. In this context, the mismatch makes the skill more suspicious because the advertised data-scraping purpose does not align with the package description at all.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High

• Category: Not specified by scanner
• Confidence: 99% confidence

Finding

The main entry point references scripts/video2text/index.js, which conflicts directly with the stated Kuaishou creator-posts/comments behavior. A mismatched executable target is dangerous because it suggests the runtime behavior may differ materially from the reviewed skill description, undermining trust boundaries and allowing hidden or unintended functionality to run when the skill is invoked. Because this is the actual execution path, it is more dangerous than a metadata-only inconsistency.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

This file implements upload/presign and AI video-analysis APIs that materially exceed the declared skill scope of Kuaishou public-data search, creator post listing, and comment collection. Scope divergence is dangerous because it can cause users or upstream agents to send unrelated local file paths, uploaded media URLs, and tokens to remote endpoints under false expectations, increasing the risk of unauthorized data handling and hidden capabilities.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

This file implements a generic local-file-to-remote-OSS upload capability, which is materially outside the declared scope of a skill that should only retrieve public Kuaishou data. In an agent environment, such functionality can be abused to exfiltrate arbitrary local files to attacker-controlled presigned URLs, and the HTTPS check does not mitigate that risk because it validates transport security, not destination trust or business purpose.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High

• Category: Not specified by scanner
• Confidence: 99% confidence

Finding

The function accepts an arbitrary filename, reads it from the local filesystem, and streams it to a caller-supplied presigned URL with caller-supplied headers. For a Kuaishou scraping skill, this is unjustified access to local files and creates a direct exfiltration primitive that could expose secrets, tokens, configuration, or other sensitive data present on the host.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

This file implements a generic video-to-text pipeline that downloads arbitrary videos, uploads them to a remote service, and retrieves AI-generated text, which is materially broader than the declared skill purpose of Kuaishou public-data search, creator post listing, and comment collection. That scope mismatch is dangerous because users or downstream agents may invoke the skill under false assumptions, causing unexpected exfiltration of non-Kuaishou content and enabling hidden data-processing capabilities not disclosed by the manifest.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The code accepts either arbitrary local file paths or arbitrary remote URLs, downloads them if needed, and uploads the resulting video to a presigned remote endpoint for analysis. In the context of a skill advertised as Kuaishou scraping, this creates an unjustified data exfiltration path for user-local files and non-Kuaishou remote content, expanding risk far beyond the expected task and potentially exposing sensitive media to third-party services.

Content

No source excerpt is available for this finding.

Whitespace Padding

Medium

• Category: Prompt Injection
• Confidence: 70% confidence

Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · README.md (reported line 99)May include surrounding context.

md

4. 同时传入文件路径与任务ID，优先执行 `--id`，忽略 `--file`
5. 无自定义 prompt 时，默认完整转录视频全部文字

| 用户自然语言指令                                         | 自动生成命令                                                                                                |
| -------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- |
| 视频提取 https://example.com/video.mp4 中的文字          | `node scripts/video2text/index.js --file "https://example.com/video.mp4"`                                   |
| 把本地 /path/to/your/video.mp4 改成小红书风格的文案      | `node scripts/video2text/index.js --file "/path/to/your/video.mp4" --prompt "改写成小红书风格的文案"`       |