← Back to archive

Repo of the Day

VoltAgent/official-mcp-servers

Published: Oct 14, 2026

Open repository ↗

A curated directory of 280+ official MCP servers from the companies behind the products. No unofficial forks, no abandoned projects

Summary

Official MCP Servers is a curated index of 301 Model Context Protocol (MCP) servers, each published by the company behind the product it connects to. The repository filters out unofficial forks and abandoned projects, organizing entries into 14 categories such as Databases, Observability, Cloud and DevOps, and Developer Tools. It works as a reference catalog rather than a runnable tool, helping engineers pick trusted servers to wire into MCP clients like Claude, Cursor, or Codex.

What it is useful for

The list solves a trust problem the README calls out: "An MCP server can read your data and act on your behalf, so who built it matters." Because MCP grants AI agents real access to databases, cloud accounts, and source repositories, choosing a vendor-maintained server instead of a community fork reduces supply-chain risk.

Engineers can use the directory to:

  • Find an official server before connecting an assistant to a sensitive system. Examples include supabase/mcp for Supabase projects, github/github-mcp-server for repos and pull requests, and mongodb-js/mongodb-mcp-server for MongoDB.
  • Locate specialized tools outside their day-to-day stack, such as grafana/mcp-grafana for incident dashboards, exa-labs/exa-mcp-server for web search, and JetBrains/mcp-jetbrains for IDE control.
  • Compare options within a category. The Databases section alone lists Postgres, MySQL, ClickHouse, Neo4j, Snowflake, Redis, and vector stores including Milvus, Qdrant, and Chroma.
  • Spot vendors whose MCP support is documented only on their docs site, since the README notes "Where there is no public repository, we link the vendor's official MCP docs" (covering Vercel, Railway, PlanetScale, and others).

How engineers can use it

This repository contains no code to install. The README is the product: a Markdown table of contents and per-category lists linking to each server's GitHub repo or vendor documentation.

A practical workflow:

  1. Open the section matching the system you want to expose, for example "Browser Automation and Web Data" to compare microsoft/playwright-mcp, ChromeDevTools/chrome-devtools-mcp, and browserbase/mcp-server-browserbase.
  2. Click through to the linked repository or vendor docs to confirm requirements and authentication.
  3. Follow that project's own instructions to register the server with your MCP client.

Because setup and credentials differ for every entry, the README provides no single install command; consult each linked project's documentation for exact steps. Treat the list as a starting point for finding candidates, not a replacement for per-server docs.