Security audit
Security checks for vulnerabilities and agentic risk
This skill mainly documents an audio-upload workflow, but its broad trigger wording and external upload of local files and API keys need user review before installation.
Review this skill before installing. Use it only when you explicitly want an agent to upload a specific audio file to api-w3stream.attoaioz.cyou and retrieve a streaming link. Do not rely on it for general audio editing, conversion, dubbing, or video processing. Use scoped or disposable API keys, verify the destination service, confirm the exact file path before upload, and avoid passing untrusted titles or paths into the shell examples.
Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.
END_POS=$((FILE_SIZE - 1))
HASH=$(md5sum /path/to/podcast.mp3 | awk '{print $1}')
curl -s -X POST "https://api-w3stream.attoaioz.cyou/api/videos/AUDIO_ID/part" \
-H 'stream-public-key: YOUR_PUBLIC_KEY' \
-H 'stream-secret-key: YOUR_SECRET_KEY' \
-H "Content-Range: bytes 0-$END_POS/$FILE_SIZE" \
The manifest description tells users to use this skill for video processing, audio editing, media conversion, and dubbing generation. The file only documents remote API calls for uploading audio and fetching stream links; it does not implement editing, conversion, dubbing, or video-processing workflows.
The skill causes transmission of local audio content and authentication headers to a third-party domain but does not prominently warn users about that data flow up front. In an agent context, insufficient disclosure can lead to accidental exfiltration of user files, metadata, and secrets to an external service the user did not explicitly approve.
The manifest description limits the skill to creating an audio object, uploading a file, completing the upload, and retrieving an HLS link. However, the capability section claims support for generic create/query/modify/delete, import/reset/export/save/convert behaviors via abstract parameters, which materially exceeds the stated upload-focused scope.
The curl example initiates transmission of API credentials and audio object metadata to a third-party service. While this is core functionality rather than hidden exfiltration, it is still a true security concern in agent environments because it moves secrets and user data off-host to an external domain.
# 领先步：创建音频对象（默认配置）
curl -s -X POST 'https://api-w3stream.attoaioz.cyou/api/videos/create' \
-H 'stream-public-key: YOUR_PUBLIC_KEY' \
-H 'stream-secret-key: YOUR_SECRET_KEY' \
-H 'Content-Type: application/json' \
This step uploads the actual local audio file plus authentication headers to an external domain. Because the skill is exec-enabled and handles local files, mistaken or automatic invocation could cause direct exfiltration of private media content and associated secrets to an unvetted third-party endpoint.
END_POS=$((FILE_SIZE - 1))
HASH=$(md5sum /path/to/podcast.mp3 | awk '{print $1}')
curl -s -X POST "https://api-w3stream.attoaioz.cyou/api/videos/AUDIO_ID/part" \
-H 'stream-public-key: YOUR_PUBLIC_KEY' \
-H 'stream-secret-key: YOUR_SECRET_KEY' \
-H "Content-Range: bytes 0-$END_POS/$FILE_SIZE" \
This code sends user-supplied content and credentials to an external third-party API. External transmission is expected for an upload skill, but it remains security-relevant because it can expose sensitive local media, associated metadata, and API secrets if invoked without explicit consent and domain trust validation.
}
# 步骤1：创建音频对象
create_resp = requests.post(f'{BASE_URL}/videos/create', headers=HEADERS, json={
'title': '我的原创音乐',
'type': 'audio'
})
L170 将技能使用条件描述为“需要视频处理、音频编辑、媒体转换、配音生成时使用”，这些都是高频、宽泛的日常任务类别，并未限定必须是“上传音频到特定流媒体平台”的场景。该描述缺少明确触发短语、边界条件或反例，可能让代理在许多并不需要上传流媒体的请求中错误调用此技能。
The examples interpolate variables such as TITLE and file paths directly into shell and JSON strings, despite warning against this practice. If any of these values are influenced by user input, shell metacharacters or quoting breaks can lead to command injection, malformed requests, or unintended file access during exec-enabled skill execution.
L003 将技能描述为“支持中文交互”，且整体文档均默认中文，没有说明是否可根据用户偏好切换语言或这是特定区域场景的必要限制。按照语言/locale 政策，若技能对语言有默认约束，最好提供用户选择或明确说明该限制的适用原因。
No suspicious patterns detected.