Add MCP Servers to Give Claude Superpowers
Out of the box, Claude Code can read files and run commands. MCP servers extend it with entirely new capabilities — databases, APIs, browser automation, and more.
// .claude/settings.json
{
"mcpServers": {
"postgres": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-postgres"],
"env": {
"DATABASE_URL": "postgresql://localhost:5432/mydb"
}
}
}
}
With that configured, Claude can now query your database directly during conversations — no copy-pasting SQL results back and forth.
Some popular MCP servers worth setting up:
- Postgres/MySQL — query databases, inspect schemas, run migrations
- GitHub — create PRs, review issues, manage repos without leaving the terminal
- Brave Search — let Claude search the web for documentation and examples
- Filesystem — give Claude controlled access to directories outside the project
You can configure them globally (~/.claude/settings.json) or per-project (.claude/settings.json). Per-project is usually better since different codebases need different tools.
# Check which MCP servers are active
claude /mcp
MCP servers are inherited by subagents too, so any parallel workers you spin up get the same capabilities.
MCP servers turn Claude from a code assistant into a fully connected development environment.
Log in to leave a comment.
Scaffold a custom MCP server with Claude to give it direct access to your internal APIs, dashboards, and microservices.
Use --strict-mcp-config to restrict Claude Code to only the MCP servers you explicitly provide, ignoring all other sources.
When an MCP server misbehaves, /mcp gives you a live view of every connected server, its status, and the tools it's exposing — right inside your session.