SuperAPI AI Directory
Partner with us中文AI newsFavorites
All toolsAI agentsazure-devops-mcp

azure-devops-mcp

The MCP server for Azure DevOps, bringing the power of Azure DevOps directly to your agents.

Open website

About this tool

Purpose

Azure DevOps MCP Server lets AI agents interact with Azure DevOps through the Model Context Protocol (MCP). It brings Azure DevOps context directly to agents, enabling them to query information such as projects, builds, repositories, test plans, teams, iterations, work items, and Wikis, and to perform operations such as creating/updating Wiki pages. The server provides a lightweight layer on top of the REST API, with high-level reasoning handled by the AI agent.

Main Capabilities

According to the materials, the server supports the following tool categories (as can be seen from the example prompts):

  • List my ADO projects
  • List the ADO builds for a specified project
  • List the ADO repositories for a specified project
  • List the test plans for a specified project
  • List the teams for a specified project
  • List the iterations for a specified project
  • List my work items in a specified project
  • List the work items for a specified project and team in the current iteration
  • List all Wikis in a specified project
  • Create a Wiki page (for example, '/Architecture/Overview' and include system design content)
  • Update a Wiki page (for example, '/Getting Started' and include new onboarding guidance)
  • Get the content of a specified page in a specified Wiki (for example, get '/API/Authentication' from the Documentation Wiki) Each tool handles one focused Azure DevOps task.

Usage

It is recommended to prioritize using the remote MCP server: it requires no installation and receives new features first. The remote server will eventually replace the local server. Use the local server only when a stdio connection is needed. The local server installation steps (using Visual Studio Code and GitHub Copilot) are as follows:

  1. Install VS Code or VS Code Insiders.
  2. Install Node.js 20 or later.
  3. Open the project in VS Code.
  4. Create .vscode/mcp.json in the project.
  5. Add the configuration (replace {organization} with your Azure DevOps organization name).
  6. Save the file and start the ado server from the MCP view in VS Code.
  7. Open GitHub Copilot Chat and switch to Agent mode.
  8. Select the Azure DevOps tool and try a prompt like "List ADO projects".
  9. When prompted, sign in with a Microsoft account that has access to the selected Azure DevOps organization. To use nightly builds, replace @azure-devops/mcp in the configuration with @azure-devops/mcp@next. For better tool selection, you can add .github/copilot-instructions.md to the project. Other supported clients include Visual Studio 2022, Codex, Claude Code, Cursor, OpenCode, and Kilo Code; see the getting started guide for details.

When to use

Suitable for scenarios where you need to let an AI agent access Azure DevOps data, such as querying projects, builds, repositories, teams, iterations, work items, and Wikis, as well as creating or updating Wiki pages. The agent can perform these operations through natural language prompts.

Limitations

  • Using the remote server is recommended; the local server will eventually be replaced, and existing local users should start planning their migration.
  • The local server includes many tools. You can load only the required tool groups by domain to avoid an overly long tool list and to help clients with tool count limits. Available domains include: core, work, work-items, search, test-plans, repositories, wiki, pipelines, advanced-security. Add -d followed by the domain name to the server arguments. Always include core so the agent can get project information. If -d is omitted, the server loads all domains.
  • You can set default Azure DevOps project and team values in .vscode/mcp.json so that tools skip selection prompts.
  • If you encounter tool issues, need support, or have feature requests, you can report them using the remote MCP server issue template. During the preview period, remote MCP server issues will be tracked through this repository.
  • The project uses the MIT license.
  • Trademark use must follow Microsoft's trademark and brand guidelines, and third-party trademarks follow their respective policies.
  • For the complete list of specific tools, see the documentation: for remote tools, see Available Tools documentation, and for local tools, see TOOLSET.md.
  • For more information on billing, pricing, platforms, and more, no details are currently provided in the materials.

Related tools

Browse category