Today we are excited to take our local MCP Server for Azure DevOps out of preview 🥳. Since the initial preview announcement, we’ve worked closely with early adopters and the community to incorporate feature suggestions and feedback. We’ve improved login and authorization, added and refined tooling, and introduced domains so users can scope active tools to stay under client limits.

🤷‍♂️ What is an MCP Server?

A local MCP Server (Model Context Provider) is a tool that sits between your AI assistant (like GitHub Copilot) and your Azure DevOps organization. Its job is to inject rich, real-time context such as work items, pull requests, test plans, and more into the prompts sent to the LLM, allowing the assistant to provide more accurate and relevant answers tailored to your specific Azure DevOps project. The Azure DevOps MCP Server supports most of the main areas of the product, including work items, wiki, repos, search, and test plans. Unlike remote or hosted MCP Servers, a local MCP Server runs inside your network or local development environment, meaning it can safely access private data and tools without that data ever leaving your system.

🤔 What does “generally available” mean?

For an open-source project like this, the difference between public preview and general availability is small. You can expect the same level of engagement from us as we continue to monitor issues, incorporate community feedback and contributions, refine existing tools, and introduce new ones. The main change is that now that the project is generally available, we will be more deliberate about how we make updates to minimize the risk of breaking existing tools and ensure greater stability for everyone.

⭐ Getting Started

Getting started is simple, though it does require a few setup steps. We recommend following the full installation and setup instructions in the repository to ensure everything is configured correctly, but here’s a quick overview to help you get started.

Navigate to the Azure DevOps MCP Server repository.

1. Open Visual Studio Code
2. Copy and paste the configuration into your local .vscode\mcp.json file.
3. Start the MCP Server and begin using the tools.

If you need additional help, check out the Getting Started and Troubleshooting guides. You can also explore various example scenarios that demonstrate how to use the different tools available in the Azure DevOps MCP Server.

Here is one of many examples:

⏭️ What is next?

We’ll continue listening to customer feedback and welcoming community contributions. In addition, we plan to revisit some of the backlog requests and invest in developing new tools to expand the project’s capabilities. As always, if you have questions, encounter issues, or want to share feedback, please create an Issue in the repository.