English | 中文文档

🚀 Empower your AI assistant with multimodal generation capabilities - A powerful AI coding assistant skill that enables image, video, speech, sound effect, 3D model, music/song generation, and web search through the 302ai command-line tool, powered by 302.AI Official Website.

💡 Cross-Tool Compatible - This Skill uses the standard SKILL.md format, supporting Claude Code, Cursor, and other compatible AI coding tools

Tell your Agent:

Install this skill: https://github.com/302ai/302ai-cli-skill

A powerful AI coding skill that lets your AI assistant understand natural language requests like "generate a watercolor image of a cat" and automatically translate them into the correct 302ai CLI commands — sending requests to 302.AI's cloud services and returning results to you.

• 🎨 Multimodal generation — Images, videos, speech, sound effects, 3D models, and music
• 🔍 Web search — Multiple search providers (Tavily, Bocha, Exa, Metaso)
• 🤖 Intelligent command translation — Automatically convert natural language into CLI commands
• ⚡ Async and sync modes — Flexible workflows for different scenarios
• 🌐 1400+ AI models — Access to 302.AI's comprehensive model ecosystem
• 🛠️ Model management — Easily list, set defaults, and query parameters

💡 Tips:

• The models above are listed for reference. For exact model IDs to use in commands, run 302ai model list <type>
• Speed, quality, and price vary across models — choose based on your actual needs. For pricing, visit the official website

You ask:

Generate a watercolor painting of a sunset over mountains

Your AI assistant executes:

# Async mode (recommended) — avoids timeout
302ai image create --prompt "watercolor painting of a sunset over mountains" --model flux-1.1-pro
302ai image fetch <taskid> --short

Result: Returns a URL to the generated image

You ask:

Make a video of a cat playing with yarn

Your AI assistant executes:

# Video generation takes 1–5 minutes, async only
302ai video create --prompt "a cat playing with yarn" --model kling-v1.6-standard
302ai video fetch <taskid> --short

Result: Returns a URL to the generated video

You ask:

Read this text aloud in a female voice: "Welcome to 302.AI"

Your AI assistant executes:

# First-time TTS setup requires cache refresh
302ai tts refresh

# Then create the speech task
302ai tts create --text "Welcome to 302.AI" --provider openai --voice alloy --model tts-1
302ai tts fetch <taskid> --short

Result: Returns a URL to the audio file

⚠️ Important: The TTS module requires running 302ai tts refresh before first use to cache the provider and voice list.

You ask:

Search for the latest news on AI developments

Your AI assistant executes:

# Default provider is Tavily (best for English content)
302ai search run --query "latest AI developments" --provider tavily

Result: Returns JSON-formatted search results

💡 Search tips:

• Default provider: tavily (best for English content)
• Chinese content: Use --provider bocha for better results
• Academic search: Use --provider metaso --category scholar or --provider exa --category "research paper"

You ask:

Create a 3D model of a coffee mug

Your AI assistant executes:

# 3D model generation takes 1–5 minutes
302ai 3d create --prompt "a coffee mug" --model hyper3d-rodin
302ai 3d fetch <taskid> --short

Result: Returns a URL to the .glb file

📦 3D file viewing: The .glb format can be opened with Blender (free), Three.js online viewer, or any 3D model viewer supporting the glTF format.

You ask:

Create a song about summer vacation

Your AI assistant executes:

# Suno generates 2 songs per task, takes 1–3 minutes
302ai song create --prompt "a cheerful song about summer vacation at the beach" --provider suno
302ai song fetch <taskid> --short

Result: Returns 2 audio URLs (Suno generates 2 variations per request)

💡 Song generation tips:

• Suno (async): Best quality, generates 2 songs per task, supports custom lyrics
• Minimax (sync): Faster, requires the --lyrics parameter
• ElevenLabs (sync): Requires --composition-plan instead of lyrics

You ask:

Write lyrics for a rock song about freedom

Your AI assistant executes:

# Generate lyrics (compatible with Suno and Minimax)
302ai song lyrics --prompt "rock song about freedom and breaking chains" --provider suno

Result: Returns formatted lyrics in JSON

📝 Using lyrics: The generated lyrics can be passed to song create --lyrics or song generate --lyrics for music generation.

You ask:

Generate music for these lyrics: [your lyrics here]

Your AI assistant executes:

# Minimax sync mode — returns the audio URL immediately
302ai song generate --lyrics "Your verse here\nChorus here\n..." --provider minimax --model speech-01-turbo

Result: Returns a single audio URL

This Skill is suitable for all AI coding tools that support the SKILL.md format.

If your tool supports claude skill commands (such as Claude Code, Cursor):

claude skill install github:302ai/302ai-cli-skill/.claude/skills/302ai-cli

For other tools that support SKILL.md:

1. Clone the 302ai-cli-skill repository or download the 302ai-cli-skill ZIP:

   git clone https://github.com/302ai/302ai-cli-skill.git

   Or click Code → Download ZIP on the 302ai/302ai-cli-skill repository page on GitHub, then extract the files.

2. Copy the 302ai-cli-skill folder to your tool's skills directory (check your tool's documentation for the location)

3. Restart your tool

When new features are released (such as the Music module), follow these steps:

# For Claude Code
claude skill update github:302ai/302ai-cli-skill/.claude/skills/302ai-cli

# For Cursor or other tools
# Re-download SKILL.md and replace the old one

pip install cli_302ai==1.0.2b2 --upgrade

⚠️ Critical: Both steps are required. Updating only the skill will not enable new modules such as 3D, Search, or Song. The CLI package version must match the skill version.

macOS / Linux:

# Add to ~/.bashrc or ~/.zshrc for persistence
export AI302_KEY="your-api-key-here"

Windows PowerShell:

# Add to PowerShell profile for persistence
$env:AI302_KEY = "your-api-key-here"

Windows CMD:

set AI302_KEY=your-api-key-here

302ai image create --prompt "a cat" --api_key "your-api-key-here"

ai302-cli/
├── SKILL.md              # Main skill instructions
├── README.md             # This file (English documentation)
├── README_CN.md          # Chinese documentation
└── references/
    ├── image.md          # Image generation commands
    ├── video.md          # Video generation commands
    ├── tts.md            # Text-to-speech commands
    ├── stt.md            # Speech-to-text commands
    ├── sfx.md            # Sound effect commands
    ├── 3d.md             # 3D model generation
    ├── song.md           # Music/song generation
    ├── search.md         # Web search
    ├── model.md          # Model management
    └── record.md         # Billing queries

Your AI assistant can chain multiple operations together:

Generate an image of a sunset, then create an animated video from it, and add background music

The assistant will:

1. Generate the image using 302ai image create
2. Use that image for 302ai video create --image
3. Generate music using 302ai song create
4. Return all three URLs

If you do not specify a model, the CLI uses smart defaults:

• Images: flux-1.1-pro (high quality)
• Videos: kling-v1.6-standard (balanced)
• TTS: Provider-specific defaults
• Songs: suno-v4 (Suno) or speech-01 (Minimax)

View available models:

302ai model list t2i    # Image models
302ai model list t2v    # Video models
302ai model list tts    # TTS models

Track your usage and costs:

# Get billing information for a specific request
302ai record get <request_id>

# Request IDs are available in the JSON output of generation commands

Detailed command options, flags, and examples for each module are available in the references/ folder:

MIT

• 302.AI Official Website
• 302.AI API Documentation
• GitHub Repository

• 📧 Email: support@302.ai
• 📖 Full documentation: doc.302.ai
• 💡 Issues and feedback: GitHub Issues