mcp-server

vMLX: Fast MLX Models

Name: vMLX: Fast MLX Models
Author: jjang-ai

Get super fast MLX models with vMLX, ideal for startup founders using MCP servers, built with Python.

intermediate⏱ 30 minutes💵 Free

838 stars84 forksPythonGuide quality 8/10 · reviewed Jun 6Updated 9/5/2026100% free · open source

What it is

Use Python to quickly make super fast machine learning models.

Real use case example

"A founder wants to add product recommendations to their e-commerce site. They use vMLX to quickly build a fast ML model, which runs on their MCP server. They train the model with a dataset of customer purchases and product features. The model helps their site suggest relevant products to each customer."

Best insideCursorClaude Desktop

When to use it

•When you need to deploy large machine learning models and want to reduce memory usage
•When you require fast model serving and inference with low latency
•When you want to survive model serving interruptions, such as restarts, with a disk cache

Ready-to-paste prompt

curl -X POST -H 'Content-Type: application/json' -d '@input.json' http://localhost:8000/predict

Heads up: Make sure you have the correct Python version installed, as vMLX has specific dependencies and may not be compatible with all Python versions

Saves to your device

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

✅ Light setup

Installs with a command or two; your AI agent can do it for you.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/vmlx && curl -fsSL https://workflowstacks.com/api/skills/vmlx/claude-skill -o ~/.claude/skills/vmlx/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

What's inside — free to inspect

No purchase needed

Read the entire source before you build — unlike paid marketplaces that hide it behind a buy button.

top-level files

folders

133.4M

source size

Apache-2.0

license

Key files

README.md

File tree

.github/

assets/

bench/

benchmarks/

docs/

native_extensions/

panel/

reports/

scripts/

tests/

vmlx_engine/

.gitignore

CHANGELOG.md

latest.json

LICENSE

pyproject.toml

pytest.ini

README.md

uv.lock

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

✅ Light setup

Installs with a command or two; your AI agent can do it for you.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/vmlx && curl -fsSL https://workflowstacks.com/api/skills/vmlx/claude-skill -o ~/.claude/skills/vmlx/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

Quick Actions

Details

Creator

jjang-ai

Language

Python

Do this with a working workflow

Pages we wrote and tested that put a skill like this to work — importable n8n templates, outcome guides and Claude Desktop configs.

MCP config

Memory (knowledge graph) MCP config

Persistent memory across chats via a local knowledge graph.

Outcome guide

Score your leads by ideal-customer fit

You describe your ideal customer once and keep leads in a Google Sheet. The workflow scores each 1–5 for fit with a one-line reason, so you sort and work the best first.