# minimax-assistant

Adopt this skill: https://proskills.md/skills/other/vipmason-minimax-assistant-skill

Page: https://proskills.md/skills/other/vipmason-minimax-assistant-skill
This document is the public skill bundle for that page. It contains SKILL.md and every helper file the catalog carries. No login is required.

## File: SKILL.md
size: 5953
upstream: https://github.com/vipmason/minimax-assistant-skill/blob/HEAD/SKILL.md

````text
---
name: minimax-assistant
description: >
  MiniMax AI Platform assistant for text, images, video, music, speech, and search.
  Use when user wants to (1) install/setup mmx CLI (from https://github.com/MiniMax-AI/cli),
  (2) generate images, music, or speech, (3) use AI chat or search,
  (4) create songs with AI-generated lyrics, or (5) check/manage MiniMax quota.
  Triggers on: "MiniMax", "mmx", "生成音乐", "生成图片", "AI生成", "音乐生成", "歌词生成".
---

# MiniMax Assistant

## Installation Flow

**IMPORTANT**: Users must install mmx CLI FIRST before using this skill.

### Step 1: Install mmx CLI

```bash
npm install -g mmx-cli
```

Repository: https://github.com/MiniMax-AI/cli

### Step 2: Authenticate

```bash
mmx auth login --api-key <YOUR_API_KEY>
```

Get API key: https://platform.minimaxi.com (CN) or https://platform.minimax.io (Global)

### Step 3: Configure

```bash
mmx config show          # View current config
mmx quota show           # View quota
mmx auth status          # Check auth status
```

## All Capabilities

| Feature | Command | Description |
|---|---|---|
| Text Chat | `mmx text chat --message "hi"` | Multi-turn conversation |
| Image Generation | `mmx image "description"` | Batch with --n, --aspect-ratio |
| Video Generation | `mmx video generate --prompt "description"` | Async with --async |
| Music Generation | `mmx music generate --prompt "style" --lyrics "lyrics"` | CLI uses music-2.5 only |
| Speech Synthesis | `mmx speech synthesize --text "text" --out audio.mp3` | 30+ voices |
| Image Understanding | `mmx vision image.jpg` | Describe images |
| Web Search | `mmx search "query"` | MiniMax search |

## Advanced: Direct API (When CLI Cannot Choose Model)

mmx CLI `music generate` hardcodes music-2.5. For music-2.6 or music-cover, use direct API:

```bash
# Music generation (supports music-2.6, music-cover)
curl -X POST "https://api.minimaxi.com/v1/music_generation" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{"model":"music-2.6","prompt":"style","lyrics":"lyrics","output_format":"hex"}'

# Lyrics generation
curl -X POST "https://api.minimaxi.com/v1/lyrics_generation" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{"mode":"write_full_song","prompt":"theme or style","title":"song title"}'
```

API Key location: `~/.mmx/config.json` -> `api_key` field

## Workflow: Lyrics to Song (Full Pipeline)

```bash
# 1. Generate lyrics
LYRICS=$(curl -s -X POST "https://api.minimaxi.com/v1/lyrics_generation" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{"mode":"write_full_song","prompt":"Chinese classical, melancholic","title":"烟雨江南"}' \
  | python3 -c "import sys,json; print(json.load(sys.stdin)['lyrics'])")

# 2. Generate song with lyrics
curl -s -X POST "https://api.minimaxi.com/v1/music_generation" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d "{\"model\":\"music-2.6\",\"prompt\":\"Chinese classical instruments, Guzheng, Pipa\",\"lyrics\":\"$LYRICS\",\"output_format\":\"hex\"}" \
  > song.json

# 3. Save audio
python3 -c "
import json
d=json.load(open('song.json'))
with open('song.mp3','wb') as f:
    f.write(bytes.fromhex(d['data']['audio']))
print('Done!')
"
```

## Scripts

- `scripts/music-gen.sh` - Music generation wrapper, supports model selection
- `scripts/lyrics-to-song.sh` - Complete lyrics to song pipeline

Usage:
```bash
./scripts/music-gen.sh --model music-2.6 --prompt "style" --lyrics "lyrics" --out /path/to/output.mp3
./scripts/lyrics-to-song.sh --prompt "theme" --title "song title" --style "instrument style" --out /path/to/output.mp3
```

## Customization

### Output Location
Use `--out` to specify any output path:
```bash
--out ~/Music/my_song.mp3
--out /tmp/song.mp3
--out ./output.mp3
```
Default: current directory with timestamp (`music_<timestamp>.mp3`)

### Streaming Mode
Add `--stream` flag to receive audio as it's generated (slower but progressive):
```bash
./scripts/music-gen.sh --stream --model music-2.6 --prompt "Jazz" --out song.mp3
```
Note: Streaming waits for complete generation but processes in chunks.

### Model Selection
Use `--model` to choose different models:
```bash
--model music-2.6        # Standard music generation (700/week)
--model music-cover      # Style transfer/cover (700/week)
```

### Region
Set via mmx CLI:
```bash
mmx config set --key region --value cn     # China
mmx config set --key region --value global # International
```

### Error Handling
- API errors are printed with status codes and messages
- Check `mmx quota show` to monitor usage
- On quota exhaustion: wait for weekly reset or upgrade plan

### Quota Display Format
**Important**: In `mmx quota show`, the format is `used/total`, not `remaining/total`:
```
music-2.6  6 / 100   # means 6 used out of 100, NOT 6 remaining
```
So if you see `100/100`, your quota is exhausted for this period.

### API Key Management
Keys are stored in `~/.mmx/config.json` (managed by mmx CLI)
```bash
mmx auth status          # Check current auth
mmx auth login          # Re-authenticate if needed
```

## Models Reference

| Model | Use Case | Notes |
|---|---|---|
| MiniMax-M* | Text Chat | Check `mmx quota show` |
| music-2.5 | Music Gen | Check `mmx quota show` |
| music-2.6 | Music Gen | Quota-based |
| music-cover | Music Cover | Quota-based |
| lyrics_generation | Lyrics Gen | Quota-based |
| image-01 | Image Gen | Quota-based |
| speech-hd | Speech | Quota-based |

Run `mmx quota show` to check your specific quota limits.

## Region Configuration

```bash
mmx config set --key region --value cn    # CN region
mmx config set --key region --value global # Global region
```

API endpoints:
- CN: `https://api.minimaxi.com`
- Global: `https://api.minimax.io`

## API Documentation

See `references/api-models.md` for detailed API reference.
````

## File: README.md
upstream: https://github.com/vipmason/minimax-assistant-skill/blob/HEAD/README.md

````text
# MiniMax Assistant Skill for OpenClaw

Enables OpenClaw AI assistants to access MiniMax AI Platform.

## Quick Start

### Step 1: Install mmx CLI

First, install the MiniMax CLI from the official repository:

```bash
npm install -g mmx-cli
```

Repository: https://github.com/MiniMax-AI/cli

### Step 2: Authenticate

```bash
mmx auth login --api-key YOUR_API_KEY
```

Get your API key at: https://platform.minimaxi.com (CN) or https://platform.minimax.io (Global)

### Step 3: Install This Skill (for OpenClaw agents)

Place the skill folder in your OpenClaw skills directory:

```
your-openclaw-workspace/skills/minimax-assistant/
├── SKILL.md
├── scripts/
│   ├── music-gen.sh          # Music generation (model selectable)
│   └── lyrics-to-song.sh    # Lyrics → Song pipeline
└── references/
    └── api-models.md         # API documentation
```

## Capabilities

| Feature | Command | Description |
|---------|---------|-------------|
| Text Chat | `mmx text chat --message "hi"` | Multi-turn conversation |
| Image Generation | `mmx image "description"` | Batch with --n |
| Video Generation | `mmx video generate --prompt "desc"` | Async with --async |
| Music Generation | `mmx music generate --prompt "style" --lyrics "lyrics"` | CLI uses music-2.5 by default |
| Speech Synthesis | `mmx speech synthesize --text "text" --out audio.mp3"` | 30+ voices |
| Image Understanding | `mmx vision image.jpg` | Describe images |
| Web Search | `mmx search "query"` | MiniMax search |

## Scripts

### music-gen.sh

Music generation with model selection support (bypasses CLI limitation).

```bash
./scripts/music-gen.sh \
  --model music-2.6 \
  --prompt "Jazz, smooth, relaxing" \
  --lyrics "[verse]\nLa la la..." \
  --out song.mp3
```

### lyrics-to-song.sh

Complete lyrics → song generation pipeline.

```bash
./scripts/lyrics-to-song.sh \
  --prompt "Chinese classical, melancholic" \
  --title "烟雨江南" \
  --style "Guzheng, Pipa" \
  --out song.mp3
```

## Available Models

| Model | Use Case | Notes |
|-------|----------|-------|
| MiniMax-M* | Text Chat | Check `mmx quota show` |
| music-2.5 | Music Gen | Check `mmx quota show` |
| music-2.6 | Music Gen | Quota-based |
| music-cover | Music Cover | Quota-based |
| lyrics_generation | Lyrics Gen | Quota-based |
| image-01 | Image Gen | Quota-based |
| speech-hd | Speech | Quota-based |

Run `mmx quota show` to check your specific quota limits.

## API Reference

See `references/api-models.md` for detailed API documentation.

## License

MIT
````
