A very easy demo which created a mcp server for on-premise Azure DevOps Server based on python
azure-devops-server-mcp-server
A minimal MCP (Model Context Protocol) server implementation for on-premises Azure DevOps Server, built with Python.
This project provides a simple bridge between LLM-based agents (via MCP) and a self-hosted Azure DevOps Server instance.
Features
This MCP server exposes the following tools:
-
List Projects
Retrieve basic information of all projects in the organization. -
Recent Work Items
Fetch recently updated work items for a specified project. -
Recent Pipelines
Retrieve recent pipeline (build) execution records and calculate success rates.
Requirements
- Python 3.9+
- Azure DevOps Server (on-premises)
- Azure DevOps Personal Access Token (PAT) with read-only permissions
Security Notice
-
This project requires an Azure DevOps Personal Access Token (PAT).
Never hard-code credentials in source code. Use environment variables instead. -
This project is intended for local, trusted environments only.
-
The MCP server exposes project metadata and work item summaries to the connected LLM.
Do not use it with untrusted or public LLM endpoints.
Disclaimer
This project is not affiliated with or endorsed by Microsoft.
VS Code MCP Configuration Example
Create a mcp.json file under .vscode/ in your workspace.
⚠️ The following configuration is for reference only.
Adjust paths according to your local environment.
{
"servers": {
"azure-devops-server": {
"type": "stdio",
"command": "<path-to-python-interpreter>",
"args": [
"<path-to-azure-devops-server-mcp-server.py>"
]
}
}
}