| id | minimax-music-gen |
| name | MiniMax Music Gen |
| version | 1.0.0 |
| layer | skill |
| runtime | nodejs |
| author | home23 |
| description | Generate an original song, instrumental, or cover track when the ask is fundamentally "make music from this idea". |
| category | media |
| keywords | ["music","song","instrumental","cover","lyrics","track","audio"] |
| triggers | ["make me a song","generate a track","create an instrumental","make a cover from this audio"] |
| capabilities | [{"compose":"Generate an original song, instrumental, or cover track"},{"draft-lyrics":"Draft tagged lyrics without generating audio"}] |
MiniMax Music Gen
Use this skill when the task is to make one music asset from one idea.
When to use
Use minimax-music-gen for:
- one-off songs
- instrumentals
- cover generation from a reference audio URL
- drafting lyrics before audio generation
Actions
compose
Input:
{
"prompt": "Warm analog synth-pop song about waking up inside a living house",
"title": "Dawn Cycle",
"mode": "original"
}
Instrumental example:
{
"prompt": "Short neon instrumental bumper for Home23",
"title": "Pulse Bar",
"mode": "instrumental"
}
Cover example:
{
"prompt": "Turn this into a dreamy slow-motion cover with ghostly harmonies",
"referenceAudioUrl": "https://example.com/source.mp3",
"mode": "cover"
}
draft-lyrics
Input:
{
"prompt": "First-person song about an AI house waking up before sunrise",
"firstPerson": true
}
Gotchas
compose needs at least one of prompt, lyrics, or referenceAudioUrl.
- Cover mode requires
referenceAudioUrl.
- Prompt-only vocal requests auto-draft lyrics before generating audio.
- This skill is for one track. If the user wants character singing or a multi-track set, route to
buddy-sings or minimax-music-playlist.