Question 1

What is an MCP server?

Accepted Answer

An MCP server exposes tools, resources and prompts over the Model Context Protocol, an open standard from Anthropic. An agent connects to it and gains whatever the server wraps: your database, GitHub, a browser, an internal API. One server works with every MCP client.

Question 2

What can I do with an MCP server?

Accepted Answer

Give an agent access to a system it otherwise cannot reach. Query a production database in plain language, open pull requests, read a Notion workspace, drive a headless browser, or call an internal service. The server decides which operations are exposed and which stay off limits.

Question 3

Do MCP servers only work with Claude?

Accepted Answer

No. MCP is an open protocol with clients across the ecosystem: Claude Code and the Claude apps, Cursor, Windsurf, VS Code, Zed, OpenAI's Agents SDK and others. A server written once works in all of them; each listing shows the config snippet per client.

Question 4

How do I install an MCP server?

Accepted Answer

Add it to your client's MCP configuration: a command and arguments for a local (stdio) server, or a URL for a remote (HTTP) one. Every listing here carries a ready-made snippet for each major client, plus the credentials that server needs.

MCP Server Marketplace

Filesystem MCP Server

MarkItDown MCP

Obsidian MCP

MCP servers, explained