ShiLiu MCP Server
Enables digital human (avatar) creation, voice cloning, and video generation through ShiLiu API, compatible with any MCP client.
README
<p align="center"> <img src="shiliu_logo.png" alt="ShiLiu Digital Human" width="320" /> </p>
ShiLiu MCP Server — Digital Human, Voice Cloning and Video Generation
An MCP server for the ShiLiu API that brings digital human (avatar) creation, voice cloning, and video generation into any MCP‑compatible client (e.g., Claude Desktop, Cline, OpenAI Agents). It exposes a focused set of tools over stdio for managing credits, listing avatars and speakers, creating avatars from image/video URLs, generating videos by text, and polling generation status.
- Features:
- Retrieve remaining credits of your account
- List avatars and speakers
- Create avatar from an image URL
- Create avatar from a video URL
- Create video by text using selected avatar and speaker
- Query generation status for avatars, speakers and videos
Requirements
- Python >= 3.12
- A ShiLiu API Key (
SHILIU_API_KEY) - Optional:
uvfor dependency management
Installation
# Recommended with uv
uv sync
# Or standard pip (editable install if you plan to develop)
pip install -e .
Configuration
Create a .env file in the project root or export environment variables:
# Required
SHILIU_API_KEY=your_api_key_here
# Optional (not strictly required by current server run mode)
MCP_HOST=127.0.0.1
MCP_PORT=8000
Get an API Key at https://shiliu.chat (Contact us).
Run Locally
You can run the server directly or via the installed console script.
# Option 1: run from source
python server.py --api-key YOUR_API_KEY
# Option 2: via entrypoint (available after install)
shiliu-mcp --api-key YOUR_API_KEY
# Option 3: module form (if installed as a package)
python -m shiliu_mcp.server --api-key YOUR_API_KEY
Notes:
- The server uses the MCP FastMCP runtime and is intended to be launched by an MCP client (stdio-based).
- If
--api-keyis omitted, ensureSHILIU_API_KEYis set in environment or.env.
Use with MCP Clients
Example for Claude Desktop (adjust paths/commands as needed):
{
"mcpServers": {
"shiliu": {
"command": "python",
"args": ["server.py"],
"env": {
"SHILIU_API_KEY": "your_api_key_here"
}
}
}
}
If installed with an executable command (shiliu-mcp):
{
"mcpServers": {
"shiliu": {
"command": "shiliu-mcp",
"args": [],
"env": {
"SHILIU_API_KEY": "your_api_key_here"
}
}
}
}
Available Tools
| Tool | Description | Params |
|---|---|---|
get_remaining_credits |
Get remaining credits | - |
get_avatar_list |
List available avatars | - |
create_avatar_by_image |
Create avatar by image URL | image_url, title? |
create_avatar_by_video |
Create avatar by video URL | video_url, title? |
get_avatar_generation_status |
Query avatar generation status | avatar_id |
get_speaker_list |
List available speakers | - |
create_speaker |
Create a new speaker (voice) | title? |
get_speaker_generation_status |
Query speaker generation status | speaker_id |
create_video_by_text |
Create video by text | avatar_id, speaker_id, text, title? |
get_video_generation_status |
Query video generation status | video_id |
Success responses use code = 0 and include relevant fields such as remaining_credits, avatars, speakers, video_id, progress, status, title, video_url, etc.
Troubleshooting
-
API Key errors
- Ensure
.envis present orSHILIU_API_KEYis exported
- Ensure
-
Network issues
- Verify connectivity to
https://api.16ai.chat
- Verify connectivity to
-
Dependency issues
- Re-install with
uv syncorpip install -e .
- Re-install with
Development Notes
- Code entry point:
server.py(CLI parses--api-key,--host,--port,--reload) - Main classes and models are in
api_client.py - Packaging metadata is defined in
pyproject.toml
Disclaimer
This project interfaces with the ShiLiu API. Please ensure you have the right to use the API and comply with the API provider's terms.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。