skills.rest
MiniMax-AI
MiniMax-AIOFFICIAL
Explore all skills by this author →
Website𝕏@MiniMax_AIGitHub

mmx-h3-video

Generate, monitor, and download MiniMax-H3 videos through the mmx CLI.

2.1k|165|Updated Mar 25, 2026
Search Tags:#video generation#text-to-video#reference images#task polling#mmx CLI#Storyboard#minimax-h3
One-click install
npx skills add https://github.com/MiniMax-AI/cli --skill mmx-h3-video
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mmx-h3-video
Source: https://github.com/MiniMax-AI/cli/tree/main/skill/h3-video
Command: npx skills add https://github.com/MiniMax-AI/cli --skill mmx-h3-video

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Creating videos with the MiniMax-H3 model through the mmx CLI involves strict credential rules, region routing, media constraints, and long-running task polling that are easy to get wrong, leading to duplicate paid tasks or failed generations.

Core Features & Use Cases

  • Multiple Generation Modes: Supports text-to-video, first/last-frame video, and multimodal reference generation with up to 9 images, 3 videos, and 3 audio files.
  • Safe Task Lifecycle Management: Uses one blocking CLI command with internal polling and download, plus strict rules preventing duplicate paid submissions after timeouts or interruptions.
  • Structured Storyboard Prompting: Expands short prompts into two-level timeline storyboards with locked continuity states for multi-reference videos.
  • Use Case: A user provides three ordered character reference images and asks for a 15-second video; the Skill builds a timestamped storyboard prompt, validates media constraints, submits one MiniMax-H3 task, waits on the same session, and downloads the finished MP4.

Quick Start

Ask the agent to generate a 15-second MiniMax-H3 video from your prompt or reference images and save the result to a local MP4 file.

Frequently Asked Questions about mmx-h3-video

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a MiniMax-H3 video from the command line?▼

Run mmx video generate with --model MiniMax-H3, a prompt, a duration of 4-15 seconds, and --download for the output file. The single blocking command submits the task, polls internally every 10 seconds, and downloads the completed MP4.

How do I use reference images and audio in MiniMax-H3 video generation?▼

Pass repeated flags such as --reference-image, --reference-video, and --reference-audio, never comma-separated paths. Limits are 9 images, 3 videos, 3 audios, and 12 total mixed items, and reference audio requires at least one reference image or video.

What API key does MiniMax-H3 video generation require?▼

MiniMax-H3 requires a Pay-as-you-go or Credit API key, not OAuth or a Token Plan Subscription key. Save it once with mmx config set and never place the literal key in generation commands; error 2013 indicates an incompatible credential.

Can I combine first-frame and reference-image modes in one MiniMax-H3 request?▼

No, frame mode and reference mode cannot be mixed in a single request. Use --image with an optional --last-frame for frame mode, or the reference flags for multimodal mode, but never both together.

What should I do when a MiniMax-H3 video task times out or fails?▼

Keep the returned taskId and never submit a duplicate paid task. For polling timeouts, report the task ID since it may still complete; for failed, cancelled, or expired statuses, report the error and get user approval before resubmitting.

When is it safe to retry the alternate region for MiniMax-H3?▼

Retry the alternate region at most once, only when the first command failed before task creation due to region detection, endpoint, or pre-submission 401/403 routing issues. Never region-retry after a taskId exists or during polling and download.

Related Skills

Explore More Design & Creative Skills→
bluewaves-creationsbluewaves-creations

veo-image-to-video

Animate a single image into a short video using Veo 3.1 via fal.ai.

Official
Basic
happycapy-aihappycapy-ai
137

video-downloader

Download videos and audio from YouTube and other platforms using yt-dlp.

Community
Intermediate
KanakMalpaniKanakMalpani

fal-ai-media

Generate images, videos, and audio via fal.ai MCP tools.

Community
Advanced
0xsline0xsline
907

video-gen

Generate AI videos via Seedance 2.0, Kling, and MiniMax Hailuo models.

Community
Advanced