cloudlabMcp

MCP server that provisions CloudLab labs/sandboxes. Tools: login, create_user, create_subscription, launch_subscription, provision_lab, config_check.

1. Setup (run once)

Requires Python 3.10+.

# from the project folder
python -m venv .venv
.\.venv\Scripts\python.exe -m pip install -r requirements.txt

(macOS/Linux: use python3 -m venv .venv and ./.venv/bin/python.)

Related MCP server: PlugLayer MCP Server

2. Configure

Edit .cursor/mcp.json and set:

• the two paths (command and args) to your absolute path to .venv\Scripts\python.exe and server.py

• the env values (CLOUDLAB_BASE_URL, CLOUDLAB_USERNAME, ... ) for your CloudLab environment.

CLOUDLAB_BASE_URL must be reachable from this machine (e.g. the machine also running CloudLab, or a reachable host).

3. Test (no editor needed)

.\.venv\Scripts\python.exe test_client.py

Expected: it prints the connected tools and your config_check values. If you see those, the server works.

4. Use in Cline / Cursor

• Add the same block from .cursor/mcp.json to your MCP config (Cline: cline_mcp_settings.json, Cursor: .cursor/mcp.json).

• Reload, confirm the server shows green with the tools.

• In Agent mode, say: "Use cloudlabMcp provision_lab to create a lab."

provision_lab runs the whole flow: login -> create user -> create lab -> launch (retries every 30s while the lab is still being created).