Nine MiniMax H3 skills: what they do, requirements, and official links
An independent guide to nine skills in the MiniMax H3 repository, from product ads to animated explainers. Learn which skill works with a file-reading agent and why the other eight require MiniMax Hub before following the official links.
A video skill can sound useful without being usable in your current agent. The official MiniMax H3 repository contains nine skills, but their environment requirements differ.
This is AI Skill Market's independent introduction, not an announcement from MiniMax. We do not distribute the skills or translated copies here. Each entry links to its upstream source.
Start with compatibility: one prompt-writing skill, eight Hub workflows
We checked the compatibility fields in the nine SKILL.md files on November 13, 2026. The h3-prompt-writing skill is documented for agents that can read local files. Writing prompts with it does not require MiniMax Hub or external API calls. It does not, however, generate a video by itself.
The other eight skills explicitly depend on MiniMax Hub and its workspace or generation tools. Putting their folders into Claude, Codex, Hermes Agent, or another agent does not provide those capabilities. The descriptions below summarize the documented workflows; we have not execution-tested their video generation.
1. h3-prompt-writing: organize instructions for H3
Use this when a video request needs clearer instructions for visuals, sound, and reference media. It covers different inputs, including text, keyframes, and reference images, clips, or audio.
You need an agent that can read the skill and its reference files. Treat prompt preparation and the separate video-generation environment as two different requirements.
2. minimalist-product-ad-generator: plan a concise product ad
This workflow develops a product-focused short from product pictures and advertising requirements. It works through copy, typography, and shot planning rather than treating the request as a single video prompt. It is a closer fit for showcasing a physical product than explaining a complex app interface.
It requires MiniMax Hub and the relevant generation features. Supply authorized images and verifiable product information.
3. 3d-animation-short-generator: develop a narrative animated short
This skill takes a story through character and setting decisions, individual shots, and assembly. Consider it when continuity across scenes matters more than producing one attractive still.
It requires MiniMax Hub, including its canvas, choice cards, and image and video tools. The workflow is designed around checkpoints in that environment.
4. papercraft-stop-motion-explainer: explain a topic with paper models
This workflow approaches an educational topic through paper characters, props, and dimensional scenes. It helps turn information into a visual explanation with a handmade look. The documented deliverables can also focus on planning, still-image ideas, or storyboards.
It requires MiniMax Hub and the corresponding generation capabilities.
5. brand-promo-video-generator: introduce a brand or service
This is a candidate for introducing a website, app, or product through a short promotional film. It organizes the message and supporting assets before production. It is not a reason to invent features or achievements.
It requires MiniMax Hub and whichever image, video, audio, or editing tools the chosen production path needs.
6. music-video-subtitle-generator: design visuals around music and lyrics
This skill considers how scenes, characters, and on-screen lyrics should relate to a track. It also addresses continuity when a longer work is assembled from several clips. It is not simply a caption-correction utility.
It requires MiniMax Hub and its generation and editing capabilities. Check the rights to the track, lyrics, and reference material separately.
7. co-op-game-intro-generator: visualize a two-player game opening
This skill builds an opening or menu-style animation around two characters and player identities. It can help present a game concept, but it does not develop a playable game.
It requires MiniMax Hub and H3 generation. Its process uses an approved image to guide the subsequent video.
8. paper-collage-explainer-generator: express an idea through collage
This workflow uses cutout-style compositions to explain information or viewpoints. Unlike the dimensional paper sets in entry four, the emphasis here is on combining collage elements and textures.
It requires MiniMax Hub image, video, and audio tools. Its default audio approach retains tactile effects rather than adding music, narration, or subtitles without a request.
9. handdrawn-live-video-generator: combine live action with drawn motion
This skill develops a short scene in which rough glowing drawings interact with a real-looking space. Contact and the timing of camera movement are part of the prompt design. It is aimed at a focused visual idea rather than a long sequence of scene changes.
It requires MiniMax Hub and H3 generation. The documented flow asks for confirmation before generating the video.
Before trying a skill
- Read its SKILL.md and reference files to check the intended task and environment.
- Follow the upstream README for installation. Having the files does not give your agent the required production tools.
- Check service availability, charges, and terms. Reading a workflow on GitHub does not mean video generation is free.
- Confirm rights to images, identities, logos, and music, and understand which material will be sent to external services.
The repository README also links to its license. Availability is not the same as unrestricted permission to use, modify, or redistribute. Check the current upstream terms before use. AI Skill Market does not host skill ZIPs or distribute translations of the skill files through this article.
Official skills and installation overview / Repository README / License linked by the official README
Checked November 13, 2026, against repository commit 199b7d0e1fd95e7b3eafa8c15eca692534142427. Upstream instructions may change.