MOSI Transcribe Diarize 多说话人转写

MOSS 多说话人转写技能。支持 URL / 本地文件 / Base64 音频输入， 输出带时间戳与 speaker 的结构化转写结果（JSON、逐段文本、按说话人汇总）。 用于会议纪要、访谈录音、多人对话整理。

What to consider before installing: - The script will upload any provided audio (including local files you point to) to https://studio.mosi.cn and sends your API key as a Bearer token. Only supply an API key you trust the service with and avoid sending highly sensitive audio unless you have an acceptable data policy. - There is a metadata inconsistency: the registry entry claims no required env vars but SKILL.md/script require MOSS_API_KEY (with fallbacks). Treat this as a documentation gap — confirm where to store/approve the key before installation. - The script requires the Python 'requests' package; the skill does not declare installation steps. Make sure your environment has required Python deps. - The code enforces HTTPS and restricts the host to studio.mosi.cn, which limits accidental exfiltration to other domains. That is good, but you should still verify the service (studio.mosi.cn) and the skill owner (currently 'unknown'). - If you need higher assurance: ask the publisher to (a) correct registry metadata to declare the required env vars, (b) provide provenance for the code (source repo, signing), and (c) document the data retention/privacy policy for uploaded audio. Rotate or scope API keys used with this skill where possible.

Type: OpenClaw Skill Name: mosi-transcribe-diarize Version: 1.0.0 The skill is a legitimate tool for audio transcription and diarization using the MOSI Studio API (studio.mosi.cn). The script `scripts/transcribe.py` includes a security validation check that restricts API requests to the official domain via HTTPS, effectively preventing the tool from being used for arbitrary data exfiltration. The logic is transparent, lacks obfuscation, and strictly follows the functionality described in the documentation.