home / mcp / mcp shouji mcp server
Provides a local, customizable MCP server for long-running feedback, offline session support, and enhanced image handling.
Configuration
View docs{
"mcpServers": {
"find-xposed-magisk-mcp-feedback": {
"command": "uv",
"args": [
"run",
"--directory",
"C:\\Users\\Administrator\\Desktop\\project\\mcp-shouji-main",
"python",
"-m",
"mcp_shouji"
],
"env": {
"MCP_DEBUG": "false",
"MCP_LANGUAGE": "zh-CN",
"MCP_WEB_HOST": "127.0.0.1",
"MCP_WEB_PORT": "8765",
"MCP_DESKTOP_MODE": "false"
}
}
}
}MCP Shouji is a local customization version that lets you run a tailored MCP server on your machine. It adds timeout control, improved image upload handling, and an offline session reconnection feature to fit offline or unstable network scenarios while keeping a smooth feedback loop.
You can run a local MCP Shouji instance by configuring the MCP client to point at your local server and starting the server process through the MCP runtime. This setup enables long-running feedback waits, broader image support, and seamless session reconnection, making it ideal for personal or development workflows where you want complete control over the environment.
Prerequisites you need before starting: ensure you have a runtime tool installed that can load and run MCP configurations (the tooling is typically invoked through a command like uv), Python installed, and a suitable local path where you will place the MCP Shouji project.
Create a local MCP configuration file with the following content. Save it as a JSON file in a location you can reference from your MCP client.
{
"mcpServers": {
"mcp_shouji_local": {
"command": "uv",
"args": [
"run",
"--directory",
"C:\\Users\\Administrator\\Desktop\\project\\mcp-shouji-main",
"python",
"-m",
"mcp_shouji"
],
"timeout": 86400,
"env": {
"MCP_DEBUG": "false",
"MCP_WEB_HOST": "127.0.0.1",
"MCP_WEB_PORT": "8765",
"MCP_DESKTOP_MODE": "false",
"MCP_LANGUAGE": "zh-CN"
},
"autoApprove": ["interactive_feedback"]
}
}
}Notes on the configuration: the server name is a local MCP Shouji instance, and the directory path should be updated to your own project location to enable full local customization. The timeout is set to 86400 seconds (24 hours) to support long-running feedback workflows, and the environment variables control debugging, web interface port, and language settings.