slmn-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@slmn-mcpis the GPU free?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
slmn
What’s inside
slmn is a small collection of independent tools, each in its own notebook/module:
module | what it does |
read, grep, and edit Jupyter notebook cells by id – no hand-rolled JSON surgery | |
read a PDF’s text; check whether a local GPU is free | |
launch and monitor background jobs on a remote host over ssh; fetch a URL; check a repo’s GitHub CI | |
automate the nbdev edit->ship loop: clean/export, a reviewed commit, push, wait for CI, and (on a feature branch) merge a PR to main | |
file-based async IPC – one side drops a message file, another picks it up (with a streaming variant) | |
the | |
an optional server exposing the same tools to any MCP client ( |
Every tool is a plain, type-hinted function. That single definition is what the CLI turns into command-line flags and what the MCP server turns into a tool schema – so the three ways to call a tool (below) never drift apart.
Related MCP server: ai-devtool-mcp-server
Installation
slmn isn’t on PyPI or conda yet, so install it from GitHub. The default install is
“lite” – just the tools and the slmn CLI:
pip install git+https://github.com/drscotthawley/slmn.gitThe MCP server is optional; add the mcp extra if you want it (it pulls in fastmcp and
friends):
pip install "slmn[mcp] @ git+https://github.com/drscotthawley/slmn.git"How to use
The same tools are reachable three ways.
Import them directly:
from slmn.nbtools import grep_nb
print(grep_nb('nbs/01_nbtools.ipynb', 'def read_nb'))From the shell, via the slmn dispatcher (run slmn with no arguments for the tool list):
slmn grep_nb nbs/01_nbtools.ipynb 'def read_nb'
slmn gpu_freeOver MCP (with the mcp extra installed), registered with any MCP-speaking client:
claude mcp add slmn -- slmn-mcpDocumentation
Full documentation is hosted on the project’s GitHub
pages; the source for each page is the
correspondingly-named notebook under nbs/.
For developers
slmn is built with nbdev: the notebooks under nbs/ are the
source of truth, and the slmn/ package is generated from them. To hack on it, install in
editable mode:
git clone https://github.com/drscotthawley/slmn.git
cd slmn
pip install -e ".[mcp]"Edit the notebooks, then regenerate the package, run the tests, and ship – all via slmn’s
own publish tool (nbdev-clean/export -> reviewed commit -> push -> wait for CI -> merge a
PR to main):
slmn publish "your commit message"from slmn.misc import gpu_free
gpu_free()2
This server cannot be deployed
Maintenance
Related MCP Connectors
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
Host your MCP tool over streamable HTTP in one command.
MCP server for progressive tool usage at any scale (see https://klavis.ai)
One MCP server exposing every tool in the Gumball portfolio.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceMCP hub server that aggregates tools from multiple domain packages into a single globally-available interface.1MIT
- FlicenseNot gradedqualityBmaintenanceMCP server for AI DevTool workflow, exposing tools and resources for code review, repository chat, and repository operations.1-
- AlicenseNot gradedqualityCmaintenanceAn MCP server that exposes a suite of developer tools and apps via a shared tool registry, enabling agents to list and call tools through the MCP protocol.MIT
- FlicenseAqualityCmaintenanceMCP server that exposes dev-tools make targets as AI-callable tools for end-to-end RHOSO-to-GPU-validated-VM bring-up.16-