Agenttools

v1.0.0Coding Agentsstable

Implements a model context protocol server using Wolfram Language

agenttoolsmcpai-integration
Share:
52
Stars
0
Downloads
0
Weekly
0/5

What is Agenttools?

Agenttools is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to implements a model context protocol server using wolfram language

Implements a model context protocol server using Wolfram Language

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Implements a model context protocol server using Wolfram Lan

Use Cases

Build MCP servers using Wolfram Language for advanced computations.
WolframResearch

Maintainer

LicenseMIT
Languagewolfram language
Versionv1.0.0
UpdatedMay 22, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx agenttools

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Agenttools

Wolfram AgentTools is a Wolfram Language toolkit that implements a full MCP server, giving AI assistants direct access to Wolfram's computational engine, natural language Wolfram|Alpha queries, semantic documentation search, and notebook read/write capabilities. It ships four predefined server profiles — Wolfram, WolframAlpha, WolframLanguage, and WolframPacletDevelopment — so you can match the tool surface to your workflow. Developers use it to run live Wolfram Language code, search the Function Repository, inspect symbol definitions, and even build and publish paclets, all from inside an MCP-capable chat client.

Prerequisites

  • Wolfram Language 14.3 or higher (Mathematica, Wolfram Engine, or Wolfram Desktop)
  • An MCP-compatible client such as Claude Desktop, Claude Code, Cursor, or VS Code
  • Optional: LLMKit subscription from Wolfram for enhanced semantic search and Wolfram|Alpha context reranking
  • Docker (optional alternative): Docker installed if you prefer not to install a local Wolfram Engine
1

Install the AgentTools paclet

Open a Wolfram Language session and install the paclet from the Wolfram Paclet Repository.

PacletInstall["Wolfram/AgentTools"]
2

Load the package

After installation, load AgentTools into your Wolfram Language session to make the installation functions available.

Needs["Wolfram`AgentTools`"]
3

Install the MCP server into your client

Run InstallMCPServer with the name of your MCP client. Each named client receives a sensible default server: Claude Desktop and Goose get the general-purpose Wolfram server; coding clients like Claude Code and Cursor get WolframLanguage. You can also pass an explicit server name as the second argument.

InstallMCPServer["ClaudeDesktop"]
(* For coding clients: *)
InstallMCPServer["ClaudeCode", "WolframLanguage"]
4

Choose a server profile (optional)

AgentTools ships four server profiles. 'Wolfram' combines computation and natural language; 'WolframAlpha' handles only natural language queries; 'WolframLanguage' adds notebook tools and code inspection; 'WolframPacletDevelopment' adds documentation and paclet release tools. Pass the profile name as the second argument to InstallMCPServer.

InstallMCPServer["ClaudeDesktop", "WolframPacletDevelopment"]
5

Restart your MCP client

Restart Claude Desktop, Claude Code, or whichever client you configured. The Wolfram tools will appear automatically in the tool list. You can verify by asking the client to list available tools.

6

(Optional) Use Docker instead of a local engine

If you prefer not to install Wolfram Language locally, pull the prebuilt Docker image. You will still need to configure your MCP client to point at the container. See the docs/docker.md file in the repository for client-specific configuration examples.

docker pull ghcr.io/wolframresearch/mcpserver:latest
7

Uninstall or update

To remove the server from a client, use UninstallMCPServer. Pass the server name as a second argument to target a specific profile.

UninstallMCPServer["ClaudeDesktop"]
(* Remove a specific profile only: *)
UninstallMCPServer["ClaudeDesktop", "Wolfram"]

Agenttools Examples

Client configuration (Claude Desktop)

After running InstallMCPServer, the config is written automatically. This is an example of what the entry looks like in claude_desktop_config.json for the WolframLanguage server profile.

{
  "mcpServers": {
    "WolframLanguage": {
      "command": "/path/to/WolframKernel",
      "args": ["-noprompt", "-run", "Needs[\"Wolfram`AgentTools`\"]; StartMCPServer[\"WolframLanguage\"]"]
    }
  }
}

Prompts to try

These prompts work in Claude Desktop or Claude Code once the server is connected. They exercise different server profiles and tool categories.

- "Evaluate Integrate[Sin[x]^2, {x, 0, Pi}] and explain the result"
- "Search the Wolfram documentation for StringMatchQ and show me examples"
- "What is the population of Tokyo according to Wolfram|Alpha?"
- "Read the notebook at ~/Documents/analysis.nb and summarize its contents"
- "Run TimeSeries[RandomReal[1, 100]] and plot a moving average with window size 10"

Troubleshooting Agenttools

InstallMCPServer completes but the tools do not appear in the client

Make sure you fully restart the client after installation — a reload is not enough. Also verify the Wolfram Engine version is 14.3 or higher by running $VersionNumber in a kernel session.

WolframAlphaContext returns errors about LLMKit

WolframAlphaContext semantic search requires an active LLMKit subscription from wolfram.com/notebook-assistant-llm-kit. Without it, the Wolfram and WolframLanguage server profiles still work for code evaluation and documentation search; only the Wolfram|Alpha semantic reranking is gated.

Docker container starts but the MCP client cannot connect

The container uses stdio transport. Your MCP client config must specify 'command': 'docker' and pass 'run -i --rm ghcr.io/wolframresearch/mcpserver:latest' in args. Check docs/docker.md in the repository for the exact JSON snippet for your client.

Frequently Asked Questions about Agenttools

What is Agenttools?

Agenttools is a Model Context Protocol (MCP) server that implements a model context protocol server using wolfram language It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Agenttools?

Follow the installation instructions on the Agenttools GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Agenttools?

Agenttools works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Agenttools free to use?

Yes, Agenttools is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "agenttools": { "command": "npx", "args": ["-y", "agenttools"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Agenttools?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides