中文 | English
A collection of Duix skills for AI agents. Each skill provides a focused digital-human workflow, with its own installation guide, usage docs, and helper scripts.
Pick the skill that matches your goal:
| Your need | Recommended skill | One-line summary |
|---|---|---|
| Make a portrait speak and chat in realtime | duix-avatar-conversation | Upload a portrait → choose a voice → get a conversation webpage link for live chat |
| Lip-sync a person video into a speaking video | duix-avatar-video-generation | Upload a person video + audio → generate a lip-synced MP4 |
💡 Still unsure? If you want “an AI digital-human webpage you can talk to”, choose conversation. If you want “a video file of a person speaking”, choose video-generation.
| Item | duix-avatar-conversation | duix-avatar-video-generation |
|---|---|---|
| Core capability | Realtime conversational digital human | Lip-synced speaking-video generation |
| Input | Portrait photo + voice + language | Person video + speech audio |
| Output | Browser conversation link (conversation_url) |
Local MP4 file path |
| Interaction | Realtime voice / text chat | Pre-generated video, non-interactive |
| Credentials | DUIX_APP_ID + DUIX_APP_KEY |
DUIX_API_KEY |
| Billing | Subscription plan (custom quota; 1 creation = 1 count) | Prepaid credits (consumed by task video duration) |
| Production time | About 20 minutes to 2 hours | A few minutes to around 20+ minutes |
| Typical use cases | Smart support, virtual teacher, brand ambassador | Product intro, weekly report video, outreach A/B tests |
| Hard gates | User must pick TTS voice + confirm charge | Asset validation + credit confirmation |
Send this prompt directly to your agent:
Please help me install the duix-avatar-conversation and duix-avatar-video-generation skills: clone https://github.com/duixcom/duix-skills into the skills directory, install duix-cli, and configure the required credentials.
# 1. Clone into your agent's skills directory
cd <your-agent-skills-path>
git clone https://github.com/duixcom/duix-skills.git
# 2. Install duix-cli
npm i duix-cli -g --registry=https://registry.npmjs.org/
# 3. Verify installation
duix-cli --version
ls duix-skills/*/README.mdAgent integration requirement: Place each skill directory under
duix-skills/into your agent's skills directory so the agent can discover*/SKILL.md.
| Item | Requirement |
|---|---|
| Node.js | ≥ 18 |
| duix-cli | Latest version (auto-checks for updates after install) |
| OS | macOS / Linux / Windows (WSL) |
| Skill | Required credentials | Where to get them |
|---|---|---|
| duix-avatar-conversation | DUIX_APP_ID + DUIX_APP_KEY |
API Key management |
| duix-avatar-video-generation | DUIX_API_KEY |
API Key management |
💰 Need to recharge?
- conversation custom quota: Pricing page
- video-generation credits: Pricing page
Upload a portrait photo, choose a voice and language, and generate an AI digital human you can chat with in the browser.
Core flow:
Portrait photo → select TTS voice (hard gate) → choose language → optional persona → confirm charge (hard gate) → generate → conversation link
Key features:
- 🎙️ 40+ languages (default English)
- 🎭 Optional persona: name, greeting, personality description
- 🔒 Dual hard gates: user must manually pick a voice and confirm billing
- 🔄 Custom quota is refunded automatically if generation fails
Input a person video and speech audio to generate a talking-head video with synced lip motion.
Core flow:
Person video + speech audio → asset validation → credit confirmation → generate → MP4 file
Key features:
- 🎬 Supports 720P ~ 4K video input
- 🎵 Supports MP3 / WAV / M4A / AAC / FLAC audio
- ⚡ Credits are consumed per task; failures may be refunded
- 📐 Auto-checks aspect ratio (16:9 or 9:16) and face detection
Skills in this repository target agent environments that support SKILL.md:
| Agent environment | Status | Notes |
|---|---|---|
| Codex | ✅ Supported | — |
| Cursor | ✅ Supported | — |
| Claude Code | ✅ Supported | — |
| Copilot | ✅ Supported | — |
| Gemini | ✅ Supported | — |
| OpenClaw | ✅ Supported | — |
Exact compatibility and prerequisites depend on each skill's README.
Q: Can I install and use both skills at the same time?
A: Yes. They share the same duix-cli, but use different credentials and billing systems, so they do not conflict.
Q: The agent cannot find the skills after installation. What should I do?
A: Confirm these files exist under the skills directory:
duix-skills/duix-avatar-conversation/SKILL.md
duix-skills/duix-avatar-conversation/scripts/duix_run.sh
duix-skills/duix-avatar-video-generation/SKILL.md
duix-skills/duix-avatar-video-generation/scripts/duix_run.sh
Then ask the agent to reload skills.
Q: What if duix-cli installation fails?
A: Make sure Node.js is ≥ 18, then install from the official registry:
npm i duix-cli -g --registry=https://registry.npmjs.org/Q: Can the two skills share the same credentials?
A: No.
duix-avatar-conversationneedsDUIX_APP_ID+DUIX_APP_KEYduix-avatar-video-generationneedsDUIX_API_KEY
Get them from the corresponding console pages.
Q: How do I check account balance?
A:
- conversation custom quota:
duix-cli avatar check - video-generation credits: see the balance-check method in each skill's docs
duix-skills/
├── README.md # This document (collection overview)
├── README_zh.md # Chinese overview
├── LICENSE # Apache License 2.0
├── duix-avatar-conversation/
│ ├── SKILL.md # Skill definition (read by agents)
│ ├── README_zh.md # Chinese usage docs
│ ├── README.md # English usage docs
│ └── scripts/
│ └── duix_run.sh # Helper script
└── duix-avatar-video-generation/
├── SKILL.md
├── README_zh.md
├── README.md
└── scripts/
└── duix_run.sh
- 💬 Email: support@duix.com
- 🐛 Bug reports: please include the skill name,
task_id(if any), full CLI output, and the steps that led to the error - 📚 Skill docs: use the links under “Skill Details” above
This repository is licensed under the Apache License 2.0.