How to bulk extract copy and assets from Xiaohongshu posts
Turn exported Xiaohongshu material into transcripts, scripts, subtitles, icon and color sets, and ready-to-post captions, all on your own Windows PC.
Skills in this bundle
Put the Xiaohongshu material you have already exported to your PC into Nexus Runner 1.0.10, then run FFVoice for transcripts, Video Content Studio for scripts and subtitles, Design Picker for icons and palettes, and Design Scribe for captions, and one pass gives you a structured, reusable asset set. Every skill runs on your own Windows machine so the files never leave it, and Nexus never signs into or operates your Xiaohongshu account: it only reads the files you hand it.
- Works on material you have already exported to your own machine. No account login, no platform scraping.
- FFVoice transcribes voiceovers offline with speaker diarization and timestamps, supporting MP3, WAV, M4A and MP4 across Chinese, English, Japanese and Korean.
- Video Content Studio produces structured output by mode: scripting, subtitle transcription, video analysis, adaptation or production planning, with top_n controlling how much comes back.
- Design Picker pulls icons, images and color palettes straight out of Figma, Sketch and Adobe XD files; Chroma Lab handles contrast checks and font pairing.
- Running all six skills once costs 24 Nex in total; the free tier gives 200 Nex per month, reset at the start of each month.
What this workflow actually does
Bulk extraction here means pulling reusable pieces out of files you already hold: what was said in a voiceover, the script and subtitles inside a video, the icons and color values inside a design file, and the copy and imagery a finished post needs. It does not crawl the platform and it does not act on your account. All six are skill-type, pay-per-run calls: Film Lens 5 Nex, FFVoice 5 Nex, Chroma Lab 5 Nex, Video Content Studio 1 Nex, Design Picker 3 Nex, Design Scribe 5 Nex. You are billed only for the ones you run.
One pass through a batch of material
How the three approaches differ
| MCP Nexus | Manual work | Generic cloud AI | |
|---|---|---|---|
| Where the files go | Skills run inside the local Runner; material never leaves the machine | Files bounce between laptops and shared drives with no clear trail | Audio and design files must be uploaded to a third-party server |
| Voiceover to text | FFVoice transcribes offline with speaker diarization and timestamps | Someone types it out by ear, replaying line by line | Usually requires an upload, with limits on length and format |
| Icons and palettes | Design Picker reads Figma, Sketch and Adobe XD files directly | A designer slices each asset and copies each color value by hand | Only sees screenshots, so layers and exact color values stay out of reach |
| Shape of the output | Structured files written to disk, ready for layout or front-end use | Scattered across chat threads, email and shared folders | Mostly chat text that still has to be turned into files by hand |
| How it is billed | Per run in Nex; all six skills total 24 Nex, free tier 200 Nex per month reset monthly | Estimated in person-days, unrelated to how much is produced | Subscription or token based, not tied to the deliverables produced |
Real use cases
FAQ
Do these skills log into or scrape my Xiaohongshu account
No. All six skills read only the files in your local folder. They do not sign into any platform account and do not operate third-party sites on your behalf, so you export the material yourself first.
Are my audio and design files uploaded to the cloud
No. The skills are invoked by Nexus Runner on your own machine, and FFVoice transcribes entirely offline with no cloud API call, so the files stay on the device throughout.
How many rounds does the free tier cover
The free tier is 200 Nex per month, reset at the start of each month, and running all six skills once costs 24 Nex, so roughly 8 full passes. Heavier use can move to Pro at 49 CNY per month with 3000 Nex, or Enterprise at 299 CNY per month with 50000 Nex.
What do I need to run it
Nexus Runner 1.0.10 on Windows 10/11 x64 only. After installing, pick the skill in the Runner and point it at your material folder; no separate Python or command line setup is required.
Which audio formats and languages are supported
MP3, WAV, M4A and MP4 among other common formats, with Chinese, English, Japanese and Korean supported. Results are segmented by speaker with timestamps, so they work directly as a subtitle draft.