Security audit
Security checks for vulnerabilities and agentic risk
This skill appears to be a disclosed cloud video/audio transcription tool, but users should understand it uploads chosen media and prompts to the provider service.
Install only if you are comfortable sending selected videos/audio, media URLs, custom prompts, and the GUAIKEI_API_TOKEN to the GuaiKei cloud service and any presigned upload storage it provides. Avoid using it for confidential media unless you trust the provider's retention and deletion claims, which are not technically enforced by the local skill package.
Undeclared token-gating logic and marketing/contact prompts are not the advertised core capability and can mislead users into providing credentials or following off-platform acquisition instructions. In a skill ecosystem, this increases phishing and trust-abuse risk, especially when combined with absent implementation of the promised transcription and content-generation functions.
Undeclared token-gating logic and marketing/contact prompts are not the advertised core capability and can mislead users into providing credentials or following off-platform acquisition instructions. In a skill ecosystem, this increases phishing and trust-abuse risk, especially when combined with absent implementation of the promised transcription and content-generation functions.
Undeclared token-gating logic and marketing/contact prompts are not the advertised core capability and can mislead users into providing credentials or following off-platform acquisition instructions. In a skill ecosystem, this increases phishing and trust-abuse risk, especially when combined with absent implementation of the promised transcription and content-generation functions.
Undeclared token-gating logic and marketing/contact prompts are not the advertised core capability and can mislead users into providing credentials or following off-platform acquisition instructions. In a skill ecosystem, this increases phishing and trust-abuse risk, especially when combined with absent implementation of the promised transcription and content-generation functions.
The README defines very broad natural-language trigger and conversion rules that map loosely phrased user requests into command execution with --file, --id, and especially free-form --prompt values. In an agent setting, this can cause the skill to be invoked for ambiguous requests or attacker-crafted phrasing, leading to unintended external processing of local files, URLs, or prior task content without a strong confirmation boundary.
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.
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"` |
Without declared permissions the skill's intent is opaque and cannot be validated.
The trigger description is extremely broad and includes many common content-creation phrases, which can cause the skill to activate in contexts where the user did not intend to send files, URLs, or prompts to this external service. Overbroad activation increases the chance of accidental data disclosure, unnecessary network use, and user confusion about what system is handling their content.
The manifest description and examples are entirely in Chinese and frame outputs around Chinese platforms and Chinese-language content styles, such as 小红书, 抖音, and 公众号文案. There is no statement offering users a language/locale choice for invocation or default behavior, which can be a language policy concern under the stated rule.
The invocation guidance instructs prioritizing this skill for broad classes of summarization, rewriting, and content extraction without sufficient constraints. In practice, that can route unrelated user tasks into a workflow that expects external media processing and token-backed network access, increasing the risk of mis-invocation and unintended third-party data transfer.
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.
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"` |
This code uses Chinese-only natural-language strings in comments, operation labels, and error messages, such as the descriptions and thrown errors, with no indication that the skill is intentionally limited to Chinese users. The policy requires flagging language or locale constraints when the skill forces a specific language without user opt-in or justification.
This code initiates HTTP/HTTPS requests to arbitrary URLs and writes the response to disk, and it also deletes files on stop/failure when configured. Although it emits internal events, there is no visible confirmation prompt, print/log disclosure, or inline warning to alert users about these potentially impactful operations.
This code sends JSON request data to a remote host and includes a TOKEN header, which can transmit user or system data off-machine. Within this file there is no confirmation prompt, user-facing disclosure, or warning comment/docstring explaining that data and credentials are being sent over the network.
The skill emits all user-facing status and warning messages in Chinese only, including error, warning, and informational text. This imposes a specific language on users without opt-in or any documented locale constraint, which matches the language/locale policy violation criteria.
The comment makes privacy and data-handling guarantees ('no leakage', 'no other use', 'auto-delete after completion') that are not enforced anywhere in this function. In a video-upload skill handling potentially sensitive user media, this creates a trust and compliance gap: users may rely on protections that do not technically exist, increasing privacy, contractual, and regulatory risk if uploaded content is retained or reused downstream.
This code presents command descriptions, help text, status messages, and errors exclusively in Chinese. That forces a specific language experience without any visible user opt-in or documented locale restriction, which matches the language/locale policy violation criteria.
This JavaScript file contains user-facing error messages, comments, and help text entirely in Chinese, including thrown errors and CLI usage output. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.
Multiple user-visible error messages are hardcoded in Chinese, including parsing, timeout, authentication, and request-failure messages. This enforces a specific language/locale without offering user choice or documenting a justified locale restriction.
This file contains natural-language documentation and user-facing error messages only in Chinese, which can constitute a language/locale policy issue when no user opt-in or region-specific scope is stated. The file does not indicate that the skill is limited to Chinese-speaking users or a China-specific deployment context.
The banner text is hard-coded in Chinese, which imposes a specific language on users. Under the policy, locale or language restrictions should either be optional for the user or clearly justified as region-specific.
No suspicious patterns detected.