Back to skill

Security audit

视频内容智能分析工具

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent cloud video transcription tool, but it needs review because it can fetch arbitrary URLs and upload media or downloaded content to external destinations under broader claims than the code enforces.

Install only if you are comfortable sending selected videos, downloaded URL contents, prompts, task IDs, filenames, and the GUAIKEI_API_TOKEN to this third-party service. Avoid giving it untrusted URLs or sensitive local videos unless the publisher clarifies upload host allowlisting, retention/deletion behavior, file size limits, and user confirmation before uploads.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (3)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/utils/validator.js:3
Finding

Unrestricted URL Download Enables Server-Side Request Forgery and Response Relay

Content
View full analysis
this.__opts.maxRedirects) { const err = new Error("Too many redirects"); this.__setState(this.__states.FAILED); this.emit("error", err); return reject(err); } const redirectedURL = /^https?:\/\//.test(response.headers.location) ? response.headers.location : new URL(response.headers.location, this.url).href; this.__isRedirected = true; this.__initProtocol(redirectedURL); this.emit("redirected", redirectedURL, this.url); return this.__start(); ``` `scripts/video2text/index.js:140-148`: ```js const presignedUrl = await video.getPresignedUrl(tokenValue, file); if (!presignedUrl || !presignedUrl?.url || presignedUrl.url === "") { throw new Error("获取预签名URL失败,请反馈给开发者"); } utils.printInfo("上传文件到安全空间..."); await upload.uploadFileToOSS(file, presignedUrl.url, presignedUrl.headers); utils.printInfo("文件上传到安全空间成功,获取视频分析任务ID"); ``` ### Technical Analysis URL validation only verifies that the input uses the HTTP or HTTPS scheme. It does not reject: - Loopback destina ...[truncated 2032 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/utils/helper.js:25
Finding

Unbounded Remote Downloads Can Exhaust Local Disk and Network Resources

Content
View full analysis
{ const options = { retry: { maxRetries: constants.CREATE_MAX_ATTEMPTS, delay: constants.RETRY_INTERVAL, }, override: { skip: true, skipSmaller: true }, }; let progressLog = ""; const dl = new Downloader(url, path, options); ``` `scripts/utils/download.js:463-469`: ```js if (!this.__isResumed) { this.__total = parseInt(response.headers["content-length"]) || null; this.__resetStats(); } if (this.__isResumed && response.statusCode === 200) { this.__isResumed = false; ``` `scripts/utils/download.js:526-565`: ```js const _fileName = this.__getFileNameFromHeaders(response.headers); this.__filePath = this.__getFilePath(_fileName); this.__fileName = this.__filePath.split(path.sep).pop(); if (fs.existsSync(this.__filePath)) { const downloadedSize = this.__getFilesizeInBytes(this.__filePath); const totalSize = this.__total ? this.__total : 0; if ( typeof this.__opts.override === "object" && this.__opts.override.skip && (this.__opts.override.skipSmaller || downloadedSize >= totalSize) ) { this.emit("skip", { totalSize: this.__total, fileName: this.__fileName, filePath: this.__filePath, downloadedSize: downloadedSize, }); this.__setState(this.__states.SKIPPED); return resolve(true); } } this.__fileStream = fs.createWriteStream(this.__filePath, {}); ``` The response is then streamed directly to the file: ```js readable.on("data", (chunk) => this.__calculateStats(chunk.length)); this.__pipes.forEach((pipe) => { readable.pipe(pipe.stream, pipe.options); readable = pipe.stre ...[truncated 1796 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/utils/upload.js:11
Finding

Remote API Can Select an Arbitrary HTTPS Upload Destination

Content
View full analysis
{ const fileStats = fs.statSync(filename); const totalSize = fileStats.size; let uploadedSize = 0; let settled = false; const fileStream = fs.createReadStream(filename); // 复制预签名 headers,并补充 OSS PUT 必需字段 const uploadHeaders = Object.assign({}, headers); // 关键修复:明确 Content-Length,否则 Node 对流式 body 默认使用 chunked // Transfer-Encoding,阿里云 OSS 预签名 PUT 通常会拒绝而返回 400/403 uploadHeaders["Content-Length"] = String(totalSize); if (!uploadHeaders["Content-Type"]) { uploadHeaders["Content-Type"] = "application/octet-stream"; } const options = { host: url.hostname, path: url.pathname + url.search, method: "PUT", headers: uploadHeaders, }; ``` The network documentation in `SKILL.md:93-96` states that requests are restricted to the service website, but the implementation accepts any HTTPS hostname supplied by the API. ### Technical Analysis The upload URL and accompanying headers are obtained from the remote API. The client validates only that the URL uses HTTPS. It does not validate the hostname, port, ...[truncated 1529 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
Findings (26)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

声明描述的是一个视频内容分析与转写技能,核心能力应围绕视频链接处理、音视频下载、ASR 转写、内容理解和文本生成展开。但提供的代码片段仅实现了通用命令行参数解析与帮助文本生成,属于基础工具函数。该代码的主要目的与声明完全不同,且没有体现任何与视频分析相关的实际行为,因此构成明显的描述与行为不一致。

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

从提供的代码看,这一文件的实际功能是一个底层下载组件:通过 http/https 请求远程 URL,处理 HEAD 获取大小、重定向、重试、超时、暂停/恢复、断点续传、文件写入、本地文件重命名与清理,并通过事件上报进度。它是一个通用 Downloader,而不是视频内容分析器。虽然声明里提到“自动下载视频”,因此下载本身可被视为整体技能的支撑细节,但当前代码片段只展示了下载能力,没有任何针对今日头条、抖音、B站等平台的视频链接解析逻辑,也没有转写、OCR、ASR、NLP 总结、翻译、场景理解等核心分析行为。故该代码片段与声明的主要用途存在明显不一致,属于材料性目的不匹配。

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description describes a full-featured video ingestion, transcription, and analysis capability. However, this code chunk implements only a small utility function that reads package.json from the local filesystem to obtain the skill/package name and caches it. That behavior is unrelated to the declared end-user purpose and does not demonstrate any of the claimed video-processing capabilities. While utility code can be supportive, this specific chunk is materially disconnected from the declared functionality, so it should be flagged as a mismatch for the supplied code chunk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

声明描述的是一个视频内容分析与转写总结类技能,但提供的代码片段实际只处理 token 合法性检查:验证字符串格式、长度与字符集;若无效则暂停技能并打印获取私有 token 的广告、网址和微信联系方式;若有效则返回 token。该行为与所宣称的视频分析、下载、ASR、翻译、文案生成等核心能力无直接对应关系。虽然这可能是配套基础设施代码,但就此代码片段本身而言,其主要行为与声明用途明显不一致,且包含未声明的商业引流/授权控制行为,因此应判定为不匹配。

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

声明描述的是一个完整的视频分析技能,核心能力应包括处理视频链接、下载媒体、语音转文字、文本清洗、内容理解和生成摘要等。而提供的代码片段只是一个工具模块,用于校验 URL 和本地文件路径是否有效。虽然这类校验可能作为整个系统的辅助实现细节存在,但就该代码片段本身而言,其实际行为与声明的核心能力明显不一致,且还包含了对本地文件系统的访问检查,这在声明中没有体现。因此应判定为描述与代码行为不匹配。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The entire README, including usage instructions, warnings, examples, and support channels, is presented only in Chinese, while also describing a general-purpose video analysis tool rather than a region-specific or compliance-constrained skill. This effectively imposes a language/locale on users without documenting an opt-in choice or justification.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README makes a strong privacy/security claim that uploaded videos are automatically destroyed immediately after processing, but later describes reuse of a historical task ID without re-uploading the source. That inconsistency suggests either retained source material or retained derived content tied to a persistent identifier, which can mislead users about data retention and expose sensitive media or transcripts longer than advertised.

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"`       |

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The documentation says videos are used only for a single parsing task, yet also advertises arbitrary follow-up prompt processing on prior task results. This creates a deceptive data-handling description: users may believe no persistent processing state exists, while the service appears to preserve reusable outputs or context that could contain sensitive content.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill advertises very broad activation triggers for common requests like summarization, extraction, rewriting, and analysis across many video-related scenarios. In an agent environment, overly broad trigger language can cause the skill to activate on loosely related user prompts, leading to unintended network uploads of user-provided media or local files to a third-party service.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The README explicitly tells the AI to prioritize this skill for a wide range of loosely bounded intents and to auto-map natural language into execution commands. That increases the chance of over-invocation and unintended data transfer, especially because the skill supports both remote URLs and local file paths and sends content to an external service.

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 · SKILL.md (reported line 121)May include surrounding context.

md
当用户用自然语言下达指令时,按以下映射生成命令,保证识别与执行一致:

| 用户自然语言指令                                     | 生成的命令                                                                                                  |
| ---------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- |
| 提取 https://example.com/video.mp4 里的文字          | `node scripts/video2text/index.js --file "https://example.com/video.mp4"`                                   |
| 总结这个视频的核心观点 https://example.com/video.mp4 | `node scripts/video2text/index.js --file "https://example.com/video.mp4" --prompt "总结这个视频的核心观点"` |

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest description and keyword list are extremely broad, covering many generic content-creation and transcription scenarios without clear boundaries on when the skill should be invoked. In an agent ecosystem, this can cause over-selection or unintended routing of user requests to a skill that downloads and processes external video links, increasing the chance of unnecessary data handling, privacy exposure, or unsafe external fetches.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The metadata is written to emphasize Chinese-language use cases and Chinese platforms, but it does not clearly state whether language behavior is automatic, user-selectable, or limited by locale. This can lead to incorrect assumptions by the orchestrator or users about output language and processing defaults, causing usability and consent issues, though it is not a direct code-execution risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This function sends the provided filename and authentication token to the /api/video/presign endpoint, which is a data-transmitting network operation. Although the code has technical docstrings, there is no confirmation prompt, user-facing log/print, or explicit warning that user file metadata and credentials are being sent to a remote API.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The getVideoId function transmits the uploaded video's URL together with the API token to /api/video/id. This is a network operation involving user/system data, but the file contains no user-facing warning, prompt, or logging that would disclose the transmission.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

This function sends prompt text and a video analysis task identifier, along with the API token, to /api/video/text. Because prompt content may contain user-provided data and the operation is a remote transmission, the absence of any user-facing disclosure or confirmation meets the missing-warning criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This code contains natural-language comments and all user-visible status/error messages in Chinese, with no indication that language is configurable or user-selected. That creates a locale-policy issue because the skill imposes a specific language on users without opt-in or documented justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This function builds and sends an HTTPS POST request containing a TOKEN header and JSON request body, which may transmit user or system data to a remote service. In this file, there is no confirmation prompt or explicit user-facing warning/log that such network transmission occurs.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The code emits all user-facing status and warning messages exclusively in Chinese, including the invalid-token warning and follow-up guidance. This enforces a specific language for users without any opt-in, fallback, or indication that the skill is intended only for a Chinese-speaking or region-specific audience.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The comment makes a strong privacy and data-lifecycle promise ('automatic deletion after transcription') that this module does not enforce or even verify. In a skill that handles user videos, this can create a real security/privacy gap: operators or downstream services may retain sensitive media longer than users expect, increasing exposure if storage is misconfigured, breached, or repurposed.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The file contains user-facing natural-language text exclusively in Chinese, including the main function comment and thrown error strings. This imposes a specific language/locale without any indication of user opt-in or a documented region-specific limitation, which matches the language-policy violation criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

This JavaScript file uses Chinese-only natural-language strings in docstrings, error messages, and generated help output, such as the module description and usage text. Under the policy rule for language/locale, this is a violation because the skill does not offer any user opt-in or configurable language choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

The code emits fixed Chinese-language error messages, including the 401/403 case, with no indication that the user can choose another language. This can violate language/locale policy when a skill imposes a specific language by default without opt-in or documented justification.

Content

No source excerpt is available for this finding.