Key Concepts
- MCP (Most Capable Person/Program): A tool or server that automates tasks, especially within the AI software development lifecycle (SDLC).
- MCP Registry: A curated list or directory of MCP servers, designed to make them discoverable for developers.
- Open Source Registry: A community-driven, open-source version of the MCP Registry, allowing for self-publication of MCP servers.
- GitHub MCP Registry: GitHub's own MCP registry, initially curated but transitioning to leverage the open-source registry.
- Self-Publication: The ability for anyone to add their MCP server to the open-source registry.
- Governance Model: A system for controlling which MCP servers are allowed within an organization, especially important for security and compliance.
- Open API: A publicly available interface definition for interacting with the MCP registry, enabling interoperability.
- Server JSON: A configuration file that defines how to run an MCP server.
- Agentic Workflows: AI-driven processes where agents (like Copilot) autonomously perform tasks.
MCP Registry: Overview and Goals
The core topic is the MCP Registry, an initiative to create a discoverable and trustworthy ecosystem of MCP servers. The goal is to move from a curated list of high-quality partner servers (currently 38 on the GitHub MCP Registry) to a community-driven registry with potentially thousands of servers. This is being achieved through an open-source registry that allows for self-publication.
Transition to Open Source and Self-Publication
The GitHub MCP Registry is transitioning to leverage an open-source registry. This will allow community members to self-publish their MCP servers. The process involves:
- Using the "Publisher" command-line tool: This tool, part of the open-source registry project, helps create a
server.jsonfile. - Initializing the server JSON: The
server.jsonfile defines how to run the MCP server. - Publishing the server: If the server has a GitHub repository, it can be published under that organization.
- Automatic propagation: Once published to the open-source registry, the server will be listed in the GitHub MCP Registry and any other downstream registries.
Quality Control and Social Signals
To maintain quality in the open-source registry, GitHub will use social signals like stars, repository ownership, and other GitHub metrics to rank and verify servers. This helps users identify reputable and well-maintained MCP servers.
Example: Markitdown, with 74,000 stars, is highlighted as a popular and trustworthy server.
GitHub MCP Server and Alternatives
The GitHub MCP server is one of the top three most used MCP servers, alongside Playwright. The GitHub MCP server currently exposes 98 tools based on the GitHub API.
Governance Model for Enterprises
A key aspect of the registry is enabling governance for enterprise users. This addresses concerns about security and compliance, allowing organizations to control which MCP servers their developers can use.
Step-by-step process:
- Admins create an allowlist: They select trusted MCP servers (e.g., GitHub MCP server, Figma, HashiCorp servers, internal servers).
- Internal registry (optional): Enterprises can create their own internal registries for private MCP servers.
- API adherence: All registries (internal and external) must adhere to the open API standard defined by the open-source registry.
- GitHub settings: Admins configure GitHub settings to use the chosen registry endpoint as the allowlist.
- Enforcement: GitHub surfaces (starting with VS Code) will enforce the allowlist, preventing the use of unapproved MCP servers.
Example: A bank might only allow MCP servers from trusted vendors and their own internal servers to prevent data exfiltration.
Open API and Standardization
The open-source registry defines an open API for interacting with MCP registries. This API includes endpoints for publishing servers and listing servers. The GitHub MCP Registry adheres to this API, as does Microsoft's API Center. This standardization allows for interoperability and enables the governance model.
Timeline and How to Contribute
While self-publication is expected to be available in a month or two, users with high-quality servers can email [email protected] for manual onboarding.
Use Case: Copilot Coding Agent and Playwright
Playwright is highlighted as a valuable MCP server, particularly for web development. It's used by the Copilot coding agent to test web applications, take screenshots, and ensure functionality. Without Playwright, the coding agent's web development capabilities would be significantly limited.
Quote: "Without Playwright, I don’t think that the coding agent would be able to do web development as well as it does." - Toby
Future Directions
The main priorities for the registry are:
- Enabling self-publication: This will empower the community to contribute and drive the registry's growth.
- Implementing the governance model: This will address enterprise concerns and unlock wider adoption of MCP.
Conclusion
The GitHub MCP Registry, powered by an open-source registry, aims to create a comprehensive and trustworthy ecosystem for MCP servers. By enabling self-publication, leveraging social signals for quality control, and providing a governance model for enterprises, the initiative seeks to democratize access to AI-powered automation tools and foster innovation within the developer community.
AI summaries can miss context or contain errors. Check important details against the original video.