◈

How to format bilibili content in bulk for creator teams

Batch-prepare bilibili covers, artwork and voice tracks with six Nexus skills that run on your own Windows machine, no file uploads.

Skills in this bundle

Quick answer

Run six Nexus skills in order inside Nexus Runner 1.0.10 on Windows 10/11 x64 — Token Forge, Brand Canvas, Film Lens, Pixel Forge, Image Shrink and Text to Speech — to turn a scattered folder of bilibili material into cover art at one fixed spec, a reusable color and type sheet, and an MP3 voice track. Images and audio are processed on your own machine and written straight back to a local folder, ready for you to upload.

  • Six skills cover four outputs: style tokens, cover artwork, licensed photography and a voice track, all running locally in Nexus Runner 1.0.10.
  • Pixel Forge and Image Shrink never upload your files; size, crop, format and compression quality are all yours to set.
  • Token Forge exports colors, typography and spacing as CSS variables, JSON or SCSS that front-end code can use directly.
  • One pass through all six skills costs 36 Nex; the free tier gives 200 Nex per month, reset at the start of each month.
  • Windows 10/11 x64 only; the skills work on your local files, and you still upload to bilibili yourself.

What this pipeline actually does

Material for one episode usually lives in several folders: screenshots at mismatched sizes, covers that look different every week, stock photos with unclear licensing, and a script that still has to be recorded. This pipeline splits that work into six repeatable skills, each doing one deterministic thing — the input is a local file or a piece of text, the output is an image, a style file or an audio file you can use as is. The skills execute on your machine and do not ship your source material out; Film Lens is the sourcing step, searching the Archive35 collection and returning license links. All six are skill-type, so you can run any one alone or chain them in the order below.

Six steps to format one episode's material

1Step 1
Lock the season's visual spec first
Use Token Forge to extract colors, typography and spacing from your design files and export them as CSS variables, JSON or SCSS. Every cover, overlay and page after this follows that one palette and type scale, so a handover does not reset the look.
2Step 2
Generate cover artwork in batches
Use Brand Canvas to generate covers and promo images from text prompts in a consistent brand style, with the style customizable. Feed it the episode's theme and get several on-brand options to lay out.
3Step 3
Source licensed photography
Use Film Lens to search Archive35's curated collection — 1,109 C2PA-verified images across 55 countries — filtering by keyword and country. It returns the matching image list, C2PA verification status and license links, so sourcing and attribution are settled in one export.
4Step 4
Normalize every image to one spec
Use Pixel Forge to resize, crop, convert format, create thumbnails, rotate, flip or grayscale in bulk, and to read metadata when you need to verify dimensions. It is deterministic Pillow-based processing: the same input gives the same output every time.
5Step 5
Compress before uploading
Use Image Shrink to compress JPEG, PNG, WebP and GIF locally, with adjustable quality and output format and no cloud upload. Run covers and in-article images through it so pages load faster while quality stays acceptable.
6Step 6
Produce the voice track
Use Text to Speech to turn the script or explainer copy into high-quality MP3 with selectable voice and speed, powered by the open-source Kokoro model running locally with no cloud API. Use it for narration, trailers or a podcast cut.

Three ways to do this

MCP NexusManual workGeneric cloud AI
Batch handlingSelect the whole folder, six skills run in sequenceOpen each file in an editor and fix it one by oneMostly one conversation at a time; you split and paste batches yourself
Where files goImages and audio processed on your machine, no uploadFiles stay local but get carried around by handSource material has to be uploaded to the provider first
Style consistencyToken Forge exports the palette and type scale; Brand Canvas renders to one styleHeld in memory and verbal convention, drifts on handoverPrompts differ each time, so results are hard to reproduce
How it is billedNex per skill run; 36 Nex for one pass through all sixBilled in people-hours; rework means reschedulingUsually a monthly seat or usage meter, not tied to a specific deliverable
What you end up withFiles on disk: images, CSS variables/JSON/SCSS, MP3Scattered across laptops and chat threadsText or a single preview in a chat window, still to be exported by hand

Real use cases

Cover art and narration ready for one episode
Feed in the episode theme and a folder of raw screenshots. Brand Canvas generates cover options, Pixel Forge normalizes everything to one size and format, Image Shrink brings the file sizes down, and Text to Speech turns the script into MP3. You end up with a consistent image set plus a narration track in a local folder.
4 skills in one pass
Freezing the visual spec for a series
After the designer exports styles, Token Forge pulls colors, typography and spacing and writes them out as CSS variables, JSON and SCSS for the front-end team. The same palette then drives Brand Canvas so every cover and overlay in the series matches, and a new teammate just follows the file.
3 export formats in one run: CSS variables, JSON, SCSS
Finding licensed photography and prepping it
Search Archive35 through Film Lens by keyword and country to get an image list, C2PA verification status and license links. Send the picks to Pixel Forge for a uniform crop and size, then to Image Shrink to cut the weight. Output is publish-ready imagery plus a traceable licensing record.
Search scope: 1,109 images across 55 countries

FAQ

How do I compress bilibili covers and article images in bulk?

Use Image Shrink for the batch. It compresses JPEG, PNG, WebP and GIF locally with adjustable quality and output format, and nothing is uploaded. If the files also need a uniform size or a WebP conversion, run Pixel Forge first, then compress.

Do these skills upload my source files anywhere?

No — image and audio processing happens entirely on your machine. Pixel Forge and Image Shrink use local libraries such as Pillow, and Text to Speech runs the open-source Kokoro model locally. Film Lens is the one search step: it queries the Archive35 collection and returns an image list with license links.

How many full passes does the free tier cover?

Five. The free tier gives 200 Nex per month, reset at the start of each month, and one pass through all six skills costs 36 Nex. For heavier use, Pro is CNY 49 per month with 3000 Nex and Enterprise is CNY 299 per month with 50000 Nex.

What do I need to run this?

Nexus Runner 1.0.10 on Windows 10/11 x64, which is the only supported platform. The skills execute inside Runner on your machine and read and write the local folders you point them at.

Can it log into bilibili and publish for me?

No. These six skills only process local files and search a licensed image library; they do not hold third-party website accounts and do not drive your screen. You upload the finished covers, images and audio yourself.

Related use cases