This is a straightforward Douyin (Chinese TikTok) video downloader that strips watermarks and extracts speech to text. You paste a share link and get back the clean video file plus a markdown transcript using SenseVoice API from SiliconFlow. The video download part works without any API key, but you'll need one if you want the speech transcription. It uses FFmpeg under the hood to rip the audio before sending it off for recognition. Outputs are organized by video ID in separate folders. The documentation is entirely in Chinese but the code structure is clear. Honestly most useful if you're doing content research or need to archive Douyin videos with their spoken content indexed as searchable text.
npx skills add https://github.com/yzfly/douyin-mcp-server --skill douyin-video