Skip to content
Use logo

Use

Verified

Fullstack MCP framework to build servers and apps for AI agents.

MCP ServerCommunicationLocal (stdio) 10.1k
View on GitHub
Updated 2026-06-15

What is the Use MCP server?

mcp-use enables developers to build MCP servers using simple APIs and to create cross-platform widgets that render in supported AI clients. It includes an inspector tool for testing and integrates with GitHub for cloud deployments with observability features.

The framework provides quickstarts for both TypeScript and Python, allowing users to define tools with schemas and attach React-based widgets for rich responses.

Install & connect

Add this to your MCP client config. Pick your client below and copy.

{
  "mcpServers": {
    "mcp-use": {
      "command": "npx",
      "args": [
        "-y",
        "create-mcp-use-app"
      ]
    }
  }
}

Package: create-mcp-use-app (npm)

Other ways to install

npx

npx
npx -y create-mcp-use-app

pip

pip
pip install mcp-use

Example prompts

Once connected, try asking your AI client:

Create a new MCP server with a weather lookup tool
How do I add a React widget to my MCP app using mcp-use?
Deploy my TypeScript MCP server to the cloud
Test my MCP server using the inspector tool

Security & permissions

Runs locally via stdio transport; requires no remote API keys but may need environment variables for cloud deployment features.

What you can do with Use

Build custom MCP servers

Define tools with Zod schemas and expose them via stdio or HTTP for AI agent use.

Create interactive widgets

Develop React components that display dynamic content like weather data inside ChatGPT or Claude.

Deploy to production

Connect a GitHub repo to Manufact MCP Cloud for hosted servers with logs and metrics.

How to use Use

  1. 1Install via npm or pip using the provided package commands.
  2. 2Run npx create-mcp-use-app@latest to scaffold a project.
  3. 3Define tools and widgets in TypeScript or Python following the examples.
  4. 4Start the server with server.listen() and test via the inspector URL.
  5. 5Deploy by linking your GitHub repository in Manufact MCP Cloud.

Use: pros & cons

Pros

  • +Supports both TypeScript and Python with unified APIs
  • +Includes built-in inspector and widget rendering
  • +Enables write-once widgets across multiple AI clients
  • +Provides cloud hosting with observability out of the box

Cons

  • Primarily example-driven documentation in the excerpt
  • Requires separate setup for production cloud features
  • Widget system tied to specific React integration
Did you find this helpful?

Frequently asked questions

It offers SDKs for both TypeScript and Python.

User reviews

Verified reviews from the community shape this listing's rating.

Loading reviews…

Sign in to review

Promote Use

Add this badge to your website, or share the tool.

DFeatured on DhanasviUse 1