MCPStack

Overview

MCPStack allows developers to stack and orchestrate multiple MCP tools into pipelines, similar to how scikit-learn works with machine learning pipelines. This approach provides both powerful functionality and security by restricting LLMs to only access tools specified in the pipeline. The project supports both CLI and programmatic workflows, with tools automatically registered through entry points and comprehensive configuration options.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• list-tools

  List all validated MCP tools available in your environment

  Try:List all available MCP toolsTry:What tools can I use with MCPStack?
• list-presets

  List all available pipeline presets

  Try:Show me all pipeline presetsTry:What presets are available?
• pipeline

  Create and manage MCP tool pipelines

  Try:Create a new pipeline with these toolsTry:Build a pipeline from MCP tools
• configure

  Configure MCP tools before adding to pipelines

  Try:Configure this MCP tool with these settingsTry:Set up this tool for my pipeline
• build

  Build a pipeline for execution in LLM environments

  Try:Build this pipeline for Claude DesktopTry:Prepare my pipeline for use
• search

  Search for MCP tools or presets by prompt query

  Try:Search for tools that can analyze dataTry:Find tools related to file processing
• run

  Execute a built pipeline with FastMCP

  Try:Run my pipeline nowTry:Execute this pipeline with FastMCP
• with_tool

  Add an MCP tool to a pipeline programmatically

  Try:Add this tool to my pipelineTry:Include this MCP tool in the workflow
• with_preset

  Use a pre-configured pipeline in Python

  Try:Use the data analysis presetTry:Run this pre-built pipeline
• with_config

  Configure the entire MCPStack instance

  Try:Configure MCPStack with these settingsTry:Set global configuration options
• with_tools

  Add multiple MCP tools to a pipeline at once

  Try:Add these tools to my pipelineTry:Include multiple MCP tools in one step
• save

  Save a built pipeline to a file

  Try:Save my pipeline configurationTry:Export this pipeline for later use

Note: Tools extracted from CLI workflow descriptions in the README. Exact tool signatures not provided in the documentation.

Comparable tools

Installation

uv add mcpstack

pip install mcpstack

uv run pre-commit install

pre-commit install

{
  "mcpServers": {
    "mcpstack": {
      "command": "uv",
      "args": ["run", "mcpstack", "run"]
    }
  }
}

FAQ

What is MCPStack?

MCPStack is a scikit-learn-like pipeline orchestrator for MCP tools, allowing you to chain multiple MCP tools together and expose them to LLM environments like Claude Desktop.

How does MCPStack enhance security?

MCPStack enhances security by allowing you to explicitly include only the tools you want an LLM to access, restricting its capabilities to only those specified in the pipeline.

Compare MCPStack with