10 MCP Servers Every Developer Should Know

The Model Context Protocol ecosystem has exploded. When Anthropic launched MCP in November 2024, there were roughly 100 servers. As of January 2026, there are over 7,800 listed in the PulseMCP directory. Every major AI company—OpenAI, Google, Microsoft—has adopted it. The Agentic AI Foundation under Linux Foundation now governs the protocol.

The problem: most MCP servers are niche, experimental, or unmaintained. Here are the ten that actually matter for everyday development work, with the exact configuration you need to get started.

1. Filesystem

What it does: Read, write, search, and manage files on your local system. This is the foundational server that turns your AI assistant into something that can actually interact with your projects.

Configuration:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"]
    }
  }
}

Why it matters: Without filesystem access, your AI can only respond to what you paste into the chat. With it, the AI can explore your codebase, read documentation, and write files directly. The path argument restricts access to a specific directory for security.

2. GitHub

What it does: Manage repositories, issues, pull requests, and search code via the GitHub API.

Configuration:

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "ghp_your_token_here"
      }
    }
  }
}

Why it matters: GitHub is where most code lives. This server enables your AI to create PRs, review code, manage issues, and search across repositories. The official server (now maintained by GitHub itself) replaced Anthropic’s original version.

3. Git

What it does: Local Git operations—clone, commit, branch, diff, log. Works with any Git repository, not just GitHub.

Configuration:

{
  "mcpServers": {
    "git": {
      "command": "uvx",
      "args": ["mcp-server-git", "--repository", "/path/to/git/repo"]
    }
  }
}

Why it matters: Complements the GitHub server by handling local repository operations. Your AI can commit changes, create branches, review diffs, and manage worktrees without pushing to remote.

4. PostgreSQL

What it does: Read-only access to PostgreSQL databases for schema inspection and queries.

Configuration (Docker recommended):

{
  "mcpServers": {
    "postgres": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "-e", "POSTGRES_URL", "mcp/postgres"],
      "env": {
        "POSTGRES_URL": "postgresql://user:pass@host:5432/mydb"
      }
    }
  }
}

Why it matters: Your AI can explore database schemas, write queries, and analyze data structures. Critical security note: always use a read-only database user and never connect production databases directly.

5. Memory (Knowledge Graph)

What it does: Persistent memory through a local knowledge graph. Entities, relations, and observations survive between sessions.

Configuration:

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-memory"],
      "env": {
        "MEMORY_FILE_PATH": "/path/to/memory.jsonl"
      }
    }
  }
}

Why it matters: AI assistants forget everything between sessions. The memory server solves this by storing facts, relationships, and context in a queryable knowledge graph. This enables your AI to remember project decisions, user preferences, and ongoing work.

6. Playwright (Browser Automation)

What it does: Browser automation using the accessibility tree—no vision models required. Navigate, click, fill forms, take screenshots.

Configuration:

{
  "mcpServers": {
    "playwright": {
      "command": "npx",
      "args": ["-y", "@playwright/mcp@latest"]
    }
  }
}

Why it matters: This is Microsoft’s official MCP server, ranked #2 globally. Unlike Puppeteer alternatives, Playwright uses the accessibility tree for deterministic interactions. It’s faster, lighter, and works across Chromium, Firefox, and WebKit.

7. Slack

What it does: Read messages, post updates, search channels, and interact with Slack workspaces.

Configuration:

{
  "mcpServers": {
    "slack": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-slack"],
      "env": {
        "SLACK_BOT_TOKEN": "xoxb-your-token",
        "SLACK_TEAM_ID": "T0123456789"
      }
    }
  }
}

Why it matters: Slack is where work communication happens. This server enables your AI to summarize channels, draft messages, search history, and stay connected to team conversations. You’ll need to create a Slack app with appropriate OAuth scopes first.

8. Notion

What it does: Read and write to Notion workspaces—databases, pages, blocks.

Configuration:

{
  "mcpServers": {
    "notion": {
      "command": "npx",
      "args": ["-y", "@notionhq/notion-mcp-server"],
      "env": {
        "NOTION_TOKEN": "ntn_****"
      }
    }
  }
}

Why it matters: If your documentation, project management, or knowledge base lives in Notion, this server connects it to your AI workflow. Notion also offers a hosted version with OAuth—no API key setup required.

9. Brave Search

What it does: Web search with privacy focus. Returns search results, local business data, images, videos, and news.

Configuration:

{
  "mcpServers": {
    "brave-search": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-brave-search"],
      "env": {
        "BRAVE_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Why it matters: Your AI assistant can search the web for current information. The free tier includes 2,000 queries per month. Sign up for an API key at brave.com/search/api.

10. Fetch

What it does: Fetch web content and convert HTML to markdown for LLM consumption.

Configuration:

{
  "mcpServers": {
    "fetch": {
      "command": "uvx",
      "args": ["mcp-server-fetch"]
    }
  }
}

Why it matters: When you need to read documentation, parse a webpage, or pull content from a URL, this server handles the conversion to AI-friendly text. Respects robots.txt by default. Simple but essential.

Getting started

Prerequisites: Node.js 18+ (required for most servers). Python 3.10+ (for Python-based servers like Git and Fetch). Docker (optional, for containerized servers like PostgreSQL).

Config file locations:

Claude Desktop (macOS): ~/Library/Application Support/Claude/claude_desktop_config.json

Claude Code: ~/.claude.json or use claude mcp add command

Cursor: ~/.cursor/mcp.json

VS Code: Use Command Palette → “MCP: Open User Configuration”

Verification: After configuring, restart your AI client. In Claude, look for the hammer icon in the chat interface—if it’s present, MCP servers are connected. Click it to see which tools are available.

Common gotchas

The most common issues are PATH problems on macOS (apps launched from Finder don’t inherit shell PATH—use absolute paths like /opt/homebrew/bin/npx) and JSON syntax errors in config files (missing commas break everything silently).

If servers fail to connect, check logs at ~/Library/Logs/Claude/mcp-server-*.log on macOS. The MCP Inspector (npx @modelcontextprotocol/inspector) is invaluable for debugging.

For more servers, browse PulseMCP (7,800+ servers), Smithery (2,200+), or the official modelcontextprotocol repository.