# QMCP MultiRoom MCP Server

A multi-room collaborative MCP server with quantum-circuit simulation, tasks, rooms, messages, blackboards, documents, plugin registration, visualizations, search, OAuth helpers, and collaborative model-training tools.

The server implementation is preserved as provided in `qmcp-server.py`. The additional deployment files are only for packaging and running it on MCPize or another container-based MCP host.

## MCPize deployment

1. Import this GitHub repository in MCPize.
2. Choose the direct deployment option.
3. MCPize will build the included `Dockerfile` and start the server over STDIO.

The default container command is:

```bash
python qmcp-server.py
```

## Local setup

Python 3.11 or newer is recommended.

```bash
python -m venv .venv
. .venv/bin/activate
pip install -r requirements.txt
python qmcp-server.py
```

The server defaults to the STDIO transport, which is the expected mode for MCPize's GitHub deployment flow.

## Optional HTTP/SSE operation

The original implementation also supports the existing command-line options and environment variables:

```bash
MCP_TRANSPORT=streamable-http MCP_HOST=0.0.0.0 MCP_PORT=8000 python qmcp-server.py
```

When a non-STDIO transport is selected, the original code starts its dashboard/OAuth web server on `WEB_PORT` or `MCP_PORT + 1`.

## Notes

- Runtime state is held in memory, matching the supplied implementation.
- The `models/` directory is created by the server when model files are used; it is intentionally not pre-populated.
- No API keys or credentials are required by the repository.
- This project is provided as-is; review the server's authentication and persistence model before exposing an HTTP endpoint publicly.