Home

MATLAB MCP Server Wiki

Welcome to the MATLAB MCP Server wiki! This server connects any AI agent to a shared MATLAB installation via the Model Context Protocol (MCP).

Installation — Prerequisites, MATLAB Engine API, server setup (Windows one-click installer, macOS/Linux, Docker)
Configuration — Full YAML config reference with environment variable overrides
MCP Tools Reference — All built-in tools: code execution, async jobs, discovery, file management, monitoring
Custom Tools — Expose your .m functions as AI-callable tools via YAML
Examples — Ready-to-run MATLAB examples: signal processing, Monte Carlo, custom libraries
Architecture — System design, elastic engine pool, async jobs, session model
Async Jobs — Long-running jobs, progress reporting, job lifecycle
Plot Conversion — MATLAB figures to interactive Plotly JSON, style preservation
Security — Workspace isolation, upload limits, function blocklist
Troubleshooting — Common issues and solutions
FAQ — Frequently asked questions

A Python MCP server that gives AI agents (Claude, Cursor, Copilot, custom agents) the ability to:

Execute MATLAB code — sync for fast commands, auto-async for long-running jobs
Get interactive plots — MATLAB figures auto-converted to interactive Plotly JSON with full style fidelity
Discover toolboxes — browse installed toolboxes, functions, and help text
Check code quality — run checkcode/mlint before execution
Manage files — upload data, read scripts, import .mat/.csv/.json/.xlsx files
Use custom libraries — expose your .m/.mex functions as first-class MCP tools with zero code changes
Scale elastically — 2–10+ MATLAB engines auto-scale based on demand
Support multiple users — per-user workspace isolation via SSE transport

Feature	Description
Execute MATLAB code	Sync for fast commands, auto-async for jobs >30s
Elastic engine pool	Scales 2–10+ engines based on demand
Interactive Plotly plots	Line, scatter, bar, area, subplots; line styles, markers, colors preserved
Async job management	Long-running simulations, progress reporting, job polling
Toolbox discovery	Browse installed toolboxes, functions, help text
Code checker	Run `checkcode`/`mlint` before execution
Multi-user (SSE)	Per-user workspace isolation, shared server
Custom tools	Expose `.m` functions as MCP tools via YAML
File I/O	Upload data, read `.m` scripts, `.mat`, `.csv`, `.json`, `.xlsx`, images
Monitoring	Pool stats, job metrics, server health, error logs
One-click Windows install	Offline `install.bat` — no admin rights, no internet
Cross-platform	Windows + macOS, MATLAB R2022b+

Platform	MATLAB Version	Transport	Notes
macOS	R2022b+	stdio, SSE	Intel & Apple Silicon
Windows	R2022b+	stdio, SSE	10/11, Python 3.10–3.12
Linux	R2022b+	stdio, SSE	With MATLAB Engine API installed
Docker	R2022b+	stdio, SSE	Mount your MATLAB installation

Windows (one-click, no admin):

git clone https://github.com/HanSur94/matlab-mcp-server-python.git
cd matlab-mcp-server-python
install.bat

macOS / Linux:

pip install matlab-mcp-python

# Single user (stdio) — simplest
matlab-mcp

# Multi-user (SSE) — shared server
matlab-mcp --transport sse

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "matlab": {
      "command": "matlab-mcp"
    }
  }
}