1. Clone the Repo

git clone https://github.com/modelcontextprotocol/servers.git
cd servers

2. Choose a Server

You can run the example servers using either npx or Docker. Each server is located under src/<server_name>. For example:

• filesystem – a code generation server that interacts with a local file system.

---

codegen.py is a code generation agent that reads your requirements and produces runnable code in a local directory.

Setup

1. Write a configuration file named server.config. This file should specify the target directory path where the output files will be written by the agent.

2. Launch the code generation agent:

python3 codegen.py

3. Speak your requirement to the agent in natural language.

Example:

"Create a FastAPI server with a single GET endpoint that returns 'Hello World'"

The generated code will be saved to the directory specified in your server.config.

---

🌐 External Tools

Some agents require API keys:

• Firecrawl (Web Search)
  Folder: src/firecrawlmpc
  Get API key: https://www.firecrawl.dev

• 21st.dev (UI Design)
  Folder: src/magicmcp
  Get API key: https://21st.dev

Add your API keys as environment variables or as instructed in each folder's README.

---

🧠 Example Projects

These are projects created by the AI code generation agent:

• nvidia/ – A tech-themed landing page
• personal_website/ – A developer portfolio site
• music_player/ – A basic web-based music player