Scientific computing MCP servers let AI agents execute numerical code, solve equations symbolically, run statistical analyses, and submit jobs to supercomputers. Instead of manually writing MATLAB scripts, R commands, or Wolfram Language expressions, researchers can have AI assistants perform these computations through the Model Context Protocol.
This review covers MCP servers for scientific computing and mathematics — numerical analysis platforms (MATLAB, NumPy), symbolic math (SymPy, Wolfram, Mathematica), statistical computing (R), scientific programming languages (Julia), HPC infrastructure (Globus), and engineering simulation (OpenFOAM). For academic paper search and bioinformatics, see our Science & Research review. For 3D simulation and digital twins, see our Digital Twins & 3D Simulation review.
The headline findings: MathWorks ships an official MATLAB MCP server (178 stars) — one of the few major scientific computing vendors to officially support MCP. RMCP provides 52 R statistical tools across 429 packages (197 stars) — the most comprehensive single-language scientific MCP server. Globus Labs connects AI to supercomputers at ALCF and NERSC. mcp.science bundles 12+ research servers under one project. Symbolic math is fragmented across many small SymPy and Wolfram wrappers.
Numerical Computing Platforms
matlab/matlab-mcp-core-server (Official MathWorks)
| Server | Stars | Language | License |
|---|---|---|---|
| matlab-mcp-core-server | 178 | MATLAB | MathWorks License |
The official MATLAB MCP server from MathWorks — not a community wrapper, but built and maintained by the MATLAB team. Released October 2025, this is one of the strongest signals of MCP adoption in scientific computing.
What Works Well
Run MATLAB through AI agents. Execute MATLAB code directly in conversations, generate scripts from natural language descriptions, and access MATLAB documentation — all through the standard MCP protocol.
Broad client support. Works with Claude Code, Visual Studio Code, GitHub Copilot, and Gemini CLI. MathWorks explicitly tests against multiple MCP clients, which is unusual attention to interoperability.
Simulink integration. Blog posts from MathWorks demonstrate simulating Simulink models through GitHub Copilot via the MCP server — meaning AI agents can interact with complex simulation workflows, not just script execution.
MathWorks also ships an MCP client. MATLAB can act as both MCP server (exposing MATLAB to AI) and MCP client (connecting MATLAB to other MCP servers). This bidirectional approach is rare in the MCP ecosystem.
What Doesn’t Work Well
Requires MATLAB license. The server needs a local MATLAB installation, which means a commercial license. This limits accessibility compared to open-source alternatives.
Not open source. Released under MathWorks’ proprietary license, not MIT or Apache-2.0. You can inspect the code on GitHub but can’t fork and modify freely.
Community MATLAB Servers
| Server | Stars | Language | License |
|---|---|---|---|
| WilliamCloudQi/matlab-mcp-server | — | Python | — |
| Tsuchijo/matlab-mcp | — | Python | — |
Two community alternatives: WilliamCloudQi’s version provides scientific computing and data analysis with natural language script generation. Tsuchijo’s version focuses on letting LLMs write and execute MATLAB scripts. Both require a local MATLAB installation but offer MIT-style openness.
Statistical Computing (R)
finite-sample/rmcp (Most Comprehensive)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| rmcp | 197 | Python | MIT | 52 |
The most comprehensive single-language scientific MCP server we’ve found. RMCP exposes 52 statistical analysis tools across 11 categories, systematically drawing from 429 R packages via CRAN task views:
- Econometrics — regression, instrumental variables, panel data
- Machine learning — classification, clustering, ensemble methods
- Time series — ARIMA, state-space models, forecasting
- Survival analysis — Kaplan-Meier, Cox proportional hazards
- Bayesian statistics — MCMC, hierarchical models
- Spatial statistics — geostatistics, point patterns
Production-ready with full MCP protocol compliance, HTTP transport and SSE support, stdio mode for Claude Desktop, and a publicly available test server. At 197 stars, it’s the most starred R MCP server.
Other R Servers
| Server | Description |
|---|---|
| Posit mcptools | Run R code in live RStudio sessions via MCP |
| chi2labs/mcpr | Expose arbitrary R functions through MCP |
| cafferychen777/Rstudio-mcp | Deep RStudio IDE integration |
| IMNMV/ClaudeR | Connect RStudio to Claude Code, Codex, Gemini via MCP |
Posit’s mcptools is notable — from the makers of RStudio and the tidyverse, it lets MCP-enabled tools interact with running R sessions. This isn’t a wrapper; it connects AI directly to your active analysis environment.
Symbolic Mathematics
sdiehl/sympy-mcp
| Server | Stars | Language | License |
|---|---|---|---|
| sympy-mcp | 41 | Python | — |
Built by Stephen Diehl (known for Haskell work), this server exposes SymPy’s full computer algebra system — symbolic manipulation, equation solving, calculus (integration, differentiation), simplification, factoring, expansion. The most focused and well-maintained SymPy MCP server.
huhabla/calculator-mcp-server (Multi-Library)
| Server | Stars | Language | License |
|---|---|---|---|
| calculator-mcp-server | — | Python | — |
Combines three Python scientific libraries in one server: SymPy for symbolic math, NumPy for numerical computation, and SciPy for statistical analysis. Includes matrix operations. A good choice if you want one server instead of three.
tufantunc/axiom-advanced-math-mcp
| Server | Stars | Language | License |
|---|---|---|---|
| axiom-advanced-math-mcp | — | TypeScript | — |
High-precision symbolic math powered by Giac/Xcas (the CAS behind HP Prime calculators) and mathjs. Covers differential equations, symbolic mathematics, and financial time-series analysis. The Giac backend provides arbitrary-precision arithmetic that SymPy-based servers typically don’t offer.
Other Math Servers
| Server | Description |
|---|---|
| EthanHenrickson/math-mcp | Basic math and statistics for LLMs |
| SHSharkar/MCP-Mathematics | 52 functions, 158 unit conversions, financial calculations |
| edwardpwtsoi/mathjs-mcp | mathjs wrapper |
| colesmcintosh/numpy-mcp | NumPy numerical computing via MCP |
| YuChenSSR/symbolica-mcp | Symbolic computing for quantum computing workflows |
Wolfram Ecosystem
paraporoco/Wolfram-MCP
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| Wolfram-MCP | 6 | Python | MIT | 11 |
The most complete Wolfram Language MCP server — 11 tools covering the core symbolic math workflow:
calculate,solve,integrate,differentiatesimplify,factor,expandmatrix_operations,statisticsexecute— arbitrary Wolfram Language code
Requires a local Wolfram Mathematica installation. Built with FastMCP.
texra-ai/mcp-server-mathematica
| Server | Stars | Language | License |
|---|---|---|---|
| mcp-server-mathematica | — | — | — |
Executes Mathematica code via wolframscript — designed for verification workflows where you want to double-check mathematical derivations. Requires Mathematica or the free Wolfram Engine for Developers.
rhenniger/MCPServer
| Server | Stars | Language |
|---|---|---|
| MCPServer | — | Wolfram Language |
An MCP server written in pure Wolfram Language — implements the protocol natively without Python or TypeScript intermediaries. Interesting architecturally, but limited adoption.
Wolfram Alpha API Servers
| Server | Language | Notes |
|---|---|---|
| StoneDot/wolframalpha-mcp-server | TypeScript | Wolfram Alpha LLM API |
| akalaric/mcp-wolframalpha | Python | Full client + server |
| cnosuke/mcp-wolfram-alpha | Go | Lightweight |
| Garoth/wolframalpha-llm-mcp | — | Structured knowledge |
| SecretiveShell/MCP-wolfram-alpha | — | Chat REPL integration |
| benhaotang/mcp-mma-docs | — | Mathematica docs only |
Five+ servers wrapping the Wolfram Alpha API — these don’t require a local Mathematica installation, just an API key. They provide computational knowledge (math, science, conversions, data) through Wolfram Alpha’s cloud service. The fragmentation is notable: none has emerged as the clear winner.
Julia
JuliaSMLM/ModelContextProtocol.jl
| Server | Language | License |
|---|---|---|
| ModelContextProtocol.jl | Julia | — |
A full MCP specification implementation in Julia — not just a wrapper, but a proper Julia package for creating and configuring MCP servers. mcp_server() is the main entry point. Useful for Julia developers who want to expose their own tools via MCP.
samtalki/AgentREPL.jl (Persistent REPL)
| Server | Language |
|---|---|
| AgentREPL.jl | Julia |
Solves Julia’s time-to-first-execution (TTFX) problem for AI workflows. The Julia process stays alive across interactions, so you only pay the compilation cost once. Provides persistent REPL sessions via MCP STDIO transport — meaning AI agents can build on previous computations within the same session.
Other Julia Servers
| Server | Description |
|---|---|
| JuliaBench/ClaudeMCPTools.jl | Basic MCP tools for Claude |
| jonathanfischer97/juliadoc-mcp | Julia documentation server |
HPC & Research Infrastructure
globus-labs/science-mcps (Supercomputing Access)
| Server | Language | License |
|---|---|---|
| science-mcps | Python | — |
From Globus Labs (University of Chicago / Argonne National Laboratory), this project connects AI agents to national supercomputing facilities:
- Globus Transfer — move data between Globus endpoints, browse remote directories, manage transfer tasks
- Globus Compute — register and execute Python functions on remote HPC endpoints (Polaris, etc.)
- Globus Search — create indices and search across Globus Search indexes
- ALCF MCP — interact with Argonne Leadership Computing Facility
- NERSC MCP — interact with the National Energy Research Scientific Computing Center
- Diaspora MCP — Diaspora Event Fabric for topic management and event streaming
A published paper (arXiv:2508.18489) demonstrates the approach: an AI agent used Globus Compute MCP to write quantum chemistry functions, execute them on Polaris, and generate HOMO-LUMO gap visualizations consistent with published literature.
pathintegral-institute/mcp.science (Research Hub)
| Server | Stars | Language | License |
|---|---|---|---|
| mcp.science | 117 | Python | MIT |
Bundles 12+ specialized MCP servers for scientific research under one umbrella. Install any server with uvx mcp-science <name>. Includes:
- GPAW DFT — density-functional-theory calculations
- Materials Project — query the Materials Project database (requires API key)
- Python sandbox — sandboxed code execution
- Jupyter — interact with Jupyter kernels
- Mathematica — Wolfram Language integration
- SSH — remote command execution
- Web fetch — HTML/PDF/text processing
- Academic search — multi-source paper search
Already covered in our Science & Research review, but its scientific computing components (DFT, Materials Project, sandboxed execution) are directly relevant here.
Engineering Simulation
OpenFOAM MCP Servers
| Server | Description |
|---|---|
| webworn/openfoam-mcp-server | CFD education with Socratic AI, mesh generation, turbulence models |
| ymg2007/openfoam-mcp | Config management, cross-platform, wind-driven rain simulation |
Two MCP servers for OpenFOAM, the open-source computational fluid dynamics (CFD) toolkit. webworn’s version is education-focused — it uses Socratic questioning and expert error resolution to teach CFD concepts. Covers OpenFOAM 12 with blockMesh mesh generation and k-ε/k-ω SST turbulence models. ymg2007’s version is practical — read and modify OpenFOAM configuration files via MCP, with specialized tools for wind-driven rain simulation.
What’s Missing
No SageMath MCP server. SageMath wraps dozens of open-source math packages (SymPy, R, Maxima, GAP, FLINT) under one interface. An MCP server would be the single widest gateway to mathematical computing.
No Octave MCP server. GNU Octave is the open-source MATLAB alternative. Given MATLAB’s commercial license requirement, an Octave MCP server would democratize access to MATLAB-compatible computing.
No dedicated SciPy MCP. SciPy appears bundled in calculator servers but has no standalone MCP server exposing its optimization, signal processing, and linear algebra capabilities.
No Maple MCP server. Maple is widely used in education and engineering for symbolic math.
No engineering simulation giants. COMSOL, ANSYS, ABAQUS, and other FEA/FEM platforms have no MCP servers. Given MathWorks’ official support, other engineering software vendors may follow.
No optimization solvers. No CPLEX, Gurobi, or CVXPY MCP servers for mathematical optimization — a major gap for operations research and engineering design.
The Verdict
Rating: 3.5 / 5 — Strong vendor signal, real ecosystems forming, but fragmented and mostly low-traffic.
The scientific computing MCP landscape has two standout features. First, official vendor support from MathWorks — the MATLAB MCP server at 178 stars is one of the most significant endorsements of MCP from a scientific computing company. Second, R’s RMCP server is genuinely comprehensive — 52 tools across 429 packages is production-grade statistical computing, not a toy wrapper.
The Julia ecosystem is also doing things right. ModelContextProtocol.jl implements the full spec natively, and AgentREPL.jl solves a real problem (TTFX) in a language-appropriate way. These are small projects but architecturally sound.
Globus Labs’ HPC access is the most novel contribution — no other MCP category lets you submit jobs to national supercomputers. The published paper demonstrating quantum chemistry computations on Polaris validates the approach.
The 3.5 rating reflects fragmentation more than quality. Symbolic math has too many small servers — at least 15 SymPy and Wolfram wrappers, none above 41 stars, with no clear winner. The Wolfram Alpha API alone has five nearly identical MCP servers. Engineering simulation is almost empty — only OpenFOAM has MCP servers, while COMSOL, ANSYS, and other industry workhorses are absent. And star counts are low across the board — even the MATLAB official server has only 178 stars, suggesting the research computing community hasn’t fully discovered MCP yet.
For researchers: start with the official MATLAB server if you have a license, RMCP for statistics, or mcp.science for multi-tool scientific workflows. For symbolic math, sdiehl/sympy-mcp is the cleanest option. For HPC, Globus Labs’ science-mcps is the only game in town — and it’s a good one.
Last updated: March 19, 2026. Star counts and features reflect what we found during research. We do not install or test these servers hands-on — our reviews are based on documentation, source code analysis, GitHub activity, and community feedback. See our methodology note for details.