Claude Marketplace's MCP Server Directory: How to Find and Install Tools Today

Anthropic has launched the Claude Marketplace, a curated platform for discovering and installing Model Context Protocol (MCP) servers. This isn't just a general app store—it's a strategic move to standardize and control the enterprise AI tooling ecosystem, starting with the tools developers use daily in Claude Code.

What Changed — A Curated Directory for MCP Servers

The Claude Marketplace, available at marketplace.anthropic.com, functions as an official directory for vetted MCP servers. These servers extend Claude Code's capabilities by connecting it to external data sources, APIs, and tools (like databases, version control, or project management systems).

Previously, finding and configuring MCP servers involved searching GitHub, reading scattered documentation, and manual configuration. The Marketplace centralizes this discovery and aims to ensure quality and security through Anthropic's curation.

What It Means For Your Claude Code Workflow

For developers using Claude Code, this translates to three immediate benefits:

1. Simplified Discovery: Find tools for specific tasks (e.g., postgres-mcp, github-mcp, jira-mcp) in one place with clear descriptions.
2. Trusted Sources: Reduce the risk of installing unvetted or malicious servers from unknown repositories.
3. Streamlined Updates: The Marketplace paves the way for easier management and updates of your installed MCP tooling.

This move by Anthropic to "control the enterprise AI economy" means they are building the definitive platform for AI-native developer tools. Using the Marketplace aligns your setup with the intended, supported ecosystem.

How To Use It Now: Browse and Install

While the Marketplace itself is a website, installation is still a CLI operation. Here’s your new workflow:

Step 1: Browse the Marketplace
Visit marketplace.anthropic.com. Browse categories or search for tools relevant to your stack (e.g., "AWS," "Linear," "Slack").

Step 2: Get the Installation Command
Each server listing provides its installation method. Most will be installed via npm (for Node.js-based servers). For example, to install a hypothetical linear-mcp server:

npm install -g @linear/mcp-server

Step 3: Configure Your claude_desktop_config.json
After installing the server package, you must add it to your Claude Code configuration. Edit your config file (typically at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS).

Add a new entry to the mcpServers object. Using the linear-mcp example with a required API key:

{
  "mcpServers": {
    "linear": {
      "command": "npx",
      "args": ["-y", "@linear/mcp-server"],
      "env": {
        "LINEAR_API_KEY": "your_linear_api_key_here"
      }
    }
    // ... your other servers
  }
}

Step 4: Restart and Use
Restart Claude Code. The new tools and context from the MCP server will be available in your coding sessions. You can ask Claude to "create a Linear issue for this bug" or "fetch my open PRs from GitHub."

Essential MCP Servers to Install First

Based on common developer needs, prioritize these from the Marketplace:

• filesystem/bash: Often built-in, but verify. Gives Claude raw access to your project files and shell.
• github-mcp: Connect to repositories, read issues, manage PRs.
• postgres-mcp: Allow Claude to query your database schema and data (safely) to inform code changes.
• figma-mcp: Import design specs and assets directly into the development context.

The Strategic Takeaway for Developers

Anthropic is betting that the best AI coding experience is a connected one. By formalizing the MCP ecosystem via the Marketplace, they're ensuring Claude Code doesn't operate in a vacuum. Your prompt to "refactor this module based on the latest design in Figma" can actually be executed because Claude has the Figma MCP server installed.

Start treating your MCP server list as a key part of your development environment. The Marketplace is now the first place to check when you think, "I wish Claude could access X." It’s how you move from a general-purpose coding assistant to a deeply integrated team member connected to all your tools.