Skip to main content
Now that you’ve learned how to run and compose MCP servers locally and remotely, connect them to Claude or CLI clients, and use them to orchestrate APIs and services, here’s a curated set of places to discover production-ready and community MCP servers you can use, extend, or remix. Use this guide to find:
  • ready-made MCP integrations for common platforms (Asana, GitHub, Supabase, etc.),
  • community examples and tutorials,
  • curated catalogs and registries,
  • and tools to bridge remote MCP endpoints to local STDIO clients.
  1. Postman Explore — MCP Generator
  • My top recommendation is the Postman Explore collection at https://postman.com/explore/MCP-generator. It hosts hundreds of published MCP server collections you can inspect, remix, and generate locally.
  • These collections are ideal for experimentation: combine parts of a public API collection (e.g., PayPal) with other services (Amadeus, Discord) to generate a single MCP server that runs on your machine. That makes it possible to orchestrate workflows (initiate a payment, post to a channel, and save a record) using natural language.
  1. OpenTools Registry — Official Integrations
  • The OpenTools registry at https://opentools.com/registry lists production-grade MCP integrations maintained by platform vendors and companies. You’ll find integrations for services such as Asana with endpoints like get_attachments, get_goals, create_goal, and more.
  • Remember: MCP provides a standardized protocol to describe and interact with APIs, object stores (S3), monitoring dashboards (Grafana), and other systems via a consistent interface.
A screenshot of a browser window open to the OpenTools registry, showing a table of Asana API endpoints and their parameter descriptions (e.g., asana_get_goals, asana_get_goal, asana_create_goal). A large white page area with a visible mouse cursor is also shown.
  1. PulseMCP Server Directory
  • Find a catalog of MCP servers, release metadata, and usage metrics at https://pulsemcp.com/servers. Each entry typically includes classification, weekly downloads, and release dates to help you choose stable or trending servers.
Screenshot of a webpage showing the MCP Server Directory with entries like "Fetch," "GitHub," "Toolbox for Databases," and "Time." Each card lists classification, estimated weekly downloads, and release date.
  1. Smithery.ai — Curated Integrations
  • https://smithery.ai offers a curated catalog of integrations with ready-to-use MCP servers (Supabase, GitHub, GitLab merge requests, weather APIs, and more). It’s a good place to find polished integrations and examples for common workflows.
A dark-themed webpage (smithery.ai) showing a catalog of integration/API cards organized under headings like "Code Repository Management" and "Weather Data APIs." Each card lists services such as GitHub, GitLab Merge Request, Weather MCP Server, and United States Weather with brief descriptions and tags.
  1. Awesome MCP Servers (GitHub)
A dark-themed browser screenshot of a GitHub README titled about MCP (Model Context Protocol), showing sections like "What is MCP?", "Clients", and "Tutorials" with links. A large white mouse cursor is visible on the left.
Table — Quick resource summary
ResourceBest forLink
Postman Explore — MCP GeneratorRapidly generate local MCP servers from API collectionshttps://postman.com/explore/MCP-generator
OpenTools registryProduction-grade, vendor-maintained integrationshttps://opentools.com/registry
PulseMCP Server DirectoryBrowse server metadata & download statshttps://pulsemcp.com/servers
Smithery.aiCurated integration catalog for common serviceshttps://smithery.ai
Awesome MCP Servers (GitHub)Community examples, clients, and tutorialshttps://github.com/punkpeye/awesome-mcp-servers

​
MCP Remote — Exposing Remote Servers to STDIO Clients

If you have a client that only supports local STDIO MCP servers (for example, certain desktop apps) but you want to connect it to a remote, authenticated MCP server, the mcp-remote bridge proxies a remote SSE/HTTP MCP endpoint to a local STDIO-style server. This lets legacy or local-only clients connect to a remote MCP endpoint as if it were running locally.
mcp-remote is very helpful when bridging remote MCP endpoints to local clients but can be experimental in some ecosystems. Pay attention to authentication, TLS, and network security when exposing remote services locally.
Example client config (JSON snippet that launches the proxy via npx): 
{
  "mcpServers": {
    "remote-example": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://remote.mcp.server/sse"
      ]
    }
  }
}
Install the mcp-remote package: 
# Install locally in your project
npm i mcp-remote

# Or install globally
npm i -g mcp-remote
Common mcp-remote CLI usages (port binding, host override, allow non-TLS): 
# Basic: proxy a remote SSE endpoint to a local port (default behavior)
npx mcp-remote https://remote.mcp.server/sse

# Bind to a specific local port (e.g., 9696)
npx mcp-remote https://remote.mcp.server/sse 9696

# Bind to a specific local host address
npx mcp-remote https://remote.mcp.server/sse --host 127.0.0.1

# Allow non-TLS (HTTP) remote endpoints (use with caution)
npx mcp-remote http://internal-service.vpc/sse --allow-http
Many desktop clients (Claude Desktop, Cursor, Windsurf, etc.) can be configured to launch such a proxy so a remote MCP server appears local to the client.

​
Anthropic and Remote MCP Servers

Anthropic’s developer docs include examples and references for remote MCP servers and how production providers typically expose MCP endpoints. A common format for remote MCP endpoints is https://mcp.<provider>.com/sse, indicating an SSE-over-HTTPS endpoint for production integrations.
A screenshot of an Anthropic developer documentation page titled "Remote MCP servers," showing a left navigation menu and main content with instructions for connecting to remote MCP servers. The page includes a blue informational notice and example server listings further down.
A browser screenshot of the Anthropic developer documentation page titled "Remote MCP server examples." It shows a table listing companies (Asana, Atlassian, Cloudflare, etc.) with descriptions and server URLs alongside a left-hand navigation menu.
  • Local MCP servers (Node, Python, Go) remain the most common environment for experimenting and developing MCP integrations. They’re easy to run, test, and hook up to local LLM clients like Claude or Cline.
  • Production usage is trending toward remote MCP servers hosted by cloud and platform vendors (Cloudflare, Atlassian, Anthropic, etc.). Remote MCP endpoints typically use SSE over HTTPS to support production-grade integrations with authentication, monitoring, and scaling.

​
Wrapping up — How to Explore Further

  • Try these resources: Postman Explore, OpenTools registry, PulseMCP, Smithery, and the GitHub awesome list.
  • Use mcp-remote to connect remote MCP endpoints to local STDIO-only clients when needed.
  • Experiment by composing multiple APIs into a single MCP server to simplify complex workflows via natural language orchestration.
Thanks for taking this lesson. If you have questions, join the course community, work through the hands-on exercises, and keep exploring MCP servers and integrations — you’ll build confidence quickly.

Watch Video