modelcontextprotocol/registry — explained in plain English
Analysis updated 2026-06-24
Browse available MCP servers to find tools that connect your AI assistant to external services like search or databases
Publish your own MCP server to the registry so other developers can discover and use it
Run the registry locally with Docker to test your MCP server integration before publishing
Build a client that queries the registry HTTP API to list and filter available servers
| modelcontextprotocol/registry | libp2p/go-libp2p | go-gost/gost | |
|---|---|---|---|
| Stars | 6,804 | 6,802 | 6,791 |
| Language | Go | Go | Go |
| Setup difficulty | moderate | moderate | moderate |
| Complexity | 3/5 | 4/5 | 3/5 |
| Audience | developer | developer | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Running locally requires Docker, publishing requires GitHub login or DNS/web domain verification.
This repository holds the MCP Registry, a directory service for MCP servers. MCP (Model Context Protocol) is a standard that lets AI assistants connect to external tools and data sources. The registry functions like an app store: developers publish their MCP servers here, and AI clients can browse and discover what is available. The analogy the README uses is apt. Just as an app store lists applications that a phone can install, the MCP Registry lists servers that an AI assistant can connect to. A server might give the AI access to a search engine, a database, a calendar, or any other external system. The registry is a web service built in Go and backed by PostgreSQL. It exposes an HTTP API that clients can query to find available servers. Developers who want to list their server can publish it using a command-line tool included in the repository. Publishing requires proving ownership, either by logging in with GitHub or by verifying that you control a domain name through a DNS or web challenge. For contributors or developers who want to run it locally, the setup uses Docker and a single make command spins up the registry along with a database. The API reached a stable v0.1 freeze in late 2025, meaning the endpoints will not have breaking changes while integrators build against it. The project is maintained by contributors from Anthropic, PulseMCP, GitHub, and Stacklok. It launched in preview in September 2025. Community discussion happens on Discord and in GitHub Discussions.
A directory service where developers can publish MCP servers, tools that let AI assistants connect to external services like search or databases, and AI clients can browse and discover what is available.
Mainly Go. The stack also includes Go, PostgreSQL, Docker.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.