WOLFRAM

Local/Self-hosted/Session based

Wolfram Local MCP

Give LLM Systems Agentic Access to Installed Wolfram Applications

Wolfram Local MCP gives agentic access to installed Wolfram applications, allowing them to power Wolfram Foundation Tool capabilities.

Using Local MCP with AI Coding Tools

Using Wolfram Local MCP, AI coding tools such as Claude Code, Cursor and Visual Studio Code can connect to installed Wolfram applications. The AI coding assistant can evaluate Wolfram Language code, search Wolfram documentation, read and write notebooks, run tests and inspect code.

Free with: Wolfram Engine Wolfram|One Mathematica

Works with: Antigravity Claude Code Claude Desktop Cline Continue Cursor LibreChat OpenCode Windsurf Zed Any client that supports MCP

Documentation

Included with desktop products

Get started immediately with Version 15 or view other access options below.

Free Trial

Access Options

Local/Self-hosted/Session based

Wolfram Local MCP

AI coding tools

Best for use with Claude Code, Cursor, etc.

Local developer environments

Keep data and execution within your environment.

Persistent sessions

Handle multistep operations, persistent data, etc.

Local file access

Work with local resources.

Notebook support

Create structured, interactive notebooks.

Included with Wolfram|One and Mathematica

Remote/Cloud/One-shot stateless

Wolfram Cloud MCP

Chat clients

Best for use with ChatGPT, Claude, etc.

Lightweight deployment

No infrastructure. Get started in seconds.

Does not support sessions

Only single requests allowed.

Does not support external files

Does not support file upload or download.

For small-scale and casual use

Remote answers only, no local interactions.

How It Works

User or agent sends query to LLM

LLM determines a need for computation and/or trusted data

Local / Self-hosted / Session based Remote / Cloud / One-shot stateless

LLM calls the Wolfram Foundation Tool/Wolfram tools via MCP, based on user setup

Wolfram returns trusted results to the LLM

LLM presents results to user or agent

The Power of Wolfram Technology

Accurate computed results, not hallucinations

With the Wolfram Foundation Tool, you're using the power of Wolfram symbolic computation to compute accurate results, free from LLM-related issues such as hallucinations.

Open-ended custom computation, not document retrieval

The Foundation Tool efficiently computes exactly what you need—whether or not it has ever been computed before—rather than trying to find a precomputed result in an existing document.

Curated Wolfram data, not web crawls

The Foundation Tool accesses the Wolfram Knowledgebase—authoritative data curated over the past two decades—not material scraped from websites.

Transparent tool input, not black-box code

Results are backed by highly readable Wolfram Language and Wolfram|Alpha input that can readily be validated by users.

Trusted Wolfram results, not AI slop

The Foundation Tool gives you access to the same Wolfram technology that's been trusted by the world's R&D leaders for more than three decades. Learn about what's new in the latest version

Wolfram Local MCP Example

See Wolfram Local MCP in action inside a coding tool like Claude Code or Cursor. Your AI coding assistant queries Wolfram Language running on your machine—no cloud roundtrip, no data leaving your machine.

Wolfram Local MCP Q&A

Getting Started

What do I need to use Wolfram Local MCP?

What is the difference between Wolfram Local MCP and Wolfram Cloud MCP?

Wolfram Local MCP allows you to get the most out of the Wolfram product you already have installed. It allows for complex, multistep interactions using Wolfram Language driven by your preferred AI harness along with access to expert-level answers using Wolfram's algorithms and knowledgebase. If you have a licensed Wolfram Engine, Mathematica or Wolfram|One, you already have access to Wolfram Local MCP. Wolfram Cloud MCP is a free service that connects your MCP-compatible application to Wolfram technology.

Do I need any additional subscriptions to use Wolfram Local MCP?

No additional subscriptions are required.

Can I use the Wolfram Local MCP with an AI client that isn't listed in the documentation?

Yes. For MCP-compatible clients not listed in the documentation, you can generate the raw JSON configuration and adapt it manually.

Troubleshooting

I'm not seeing Wolfram tools in my chat client after following the documentation. What do I do now?

  • Fully restart your chat client after installing—closing the window often just minimizes it to the system tray.
  • Inspect the configuration of MCP servers for your client. If you want to see what Wolfram knows about a given client, you can use PacletSymbol["Wolfram/AgentTools", "DeployedAgentTools"][].
  • Check your client's documentation for the location of log files and check for errors.
  • If none of these solve the problem, you can visit the Support Center for more assistance.

Why did my computation time out?

The default setting for evaluations is 60 seconds. For long-running computations, you can ask the AI to increase the timeout setting or break your computation into smaller steps.

Why is Wolfram Local MCP using a different version of Wolfram Language?

The installed Wolfram Local MCP will use the same version of Wolfram Language as the session it was installed from. If you want to use a different version of Wolfram Language, you need to install Local MCP in a session of that version or manually edit the configuration file to point to a different Wolfram kernel.

Why aren't the results I'm getting closer to what I'm looking for?

You can review the Writing Persistent Instructions and Tips for Best Results sections of the Quick Start for Chat Clients. You can also check out Best Practices for AI-Assisted Development.

Wolfram Enterprise MCP

All-in-one solution for enterprise deployment of Wolfram Foundation Tool technology.

Contact Us