Azure DevOps Remote MCP Server
Official Azure DevOps MCP Server. Read and manage work items, repositories, pull requests, wiki, builds, releases, and project resources.
Documentation
Enable AI assistance with the Azure DevOps MCP Server - Azure Boards | Microsoft Learn
Azure DevOps Services
Consider asking your AI assistant "Get my current sprint work items, then identify which ones might be at risk" and getting instant access to your actual Azure DevOps data. The Azure DevOps Model Context Protocol (MCP) Server provides your AI assistant with secure access to work items, pull requests, builds, test plans, and documentation from your Azure DevOps organization.
Azure DevOps provides remote and local MCP Servers. The remote server is hosted and managed by Azure DevOps. The local server runs on your machine and supports clients that can't authenticate to the remote server.
Important
- The Azure DevOps MCP Server is free to use. However, standard Azure DevOps pricing applies to your organization and any data access through the service. AI assistant usage might have separate costs depending on your chosen AI platform.
- The Azure DevOps MCP Server requires your AI assistant to operate in agent-mode to access Azure DevOps data and perform operations.
Choose remote or local
Use the remote Azure DevOps MCP Server when your environment supports it. The remote server is the recommended option because it requires no local installation and Azure DevOps manages its updates. Supported environments include Visual Studio Code, Visual Studio, Microsoft Foundry, Microsoft Copilot Studio, and GitHub Copilot.
Use the local server for clients such as Claude Desktop, Claude Code, Cursor, and Codex that can't authenticate to the remote server with Microsoft Entra ID.
Install the local Azure DevOps MCP Server
Prerequisites
System requirements:Node.js 20.0+ and an active Azure DevOps organization.
The Azure DevOps MCP Server integrates with various development environments and AI assistants. Choose your preferred environment for instructions. The prerequisites listed in the table are environment-specific requirements in addition to the system requirements previously listed.
| Environment | Prerequisites | Installation | Features |
|---|---|---|---|
| Visual Studio Code (recommended) | GitHub Copilot or Claude Dev extension | One-click install | Extensive MCP support with multiple AI assistant options |
| GitHub Copilot CLI | GitHub Copilot CLI installed and authenticated | GitHub Copilot CLI setup guide | MCP-enabled Azure DevOps access from GitHub Copilot CLI |
| Visual Studio (2022 and later) | GitHub Copilot component | Visual Studio setup guide | Full IntelliSense integration with Azure DevOps data |
| Cursor | Built-in AI assistant (no extensions needed) | Cursor setup guide | Native MCP integration |
| Claude Desktop | Claude Desktop application | Claude Desktop setup guide | Standalone application with full Azure DevOps integration |
| JetBrains IDEs | Compatible AI assistant plugin | JetBrains setup guide | IDE-specific integration through plugins |
| Other environments | Varies by environment | Azure DevOps MCP Server docs repository | See repository for all options |
Tip
Having installation issues? Check the troubleshooting section or report issues on the Azure DevOps MCP Server GitHub repository.
Why use Azure DevOps MCP Server?
Traditional AI assistants lack context about your specific projects, work items, and team processes. They can help with generic coding questions but can't answer "What's blocking our current sprint?" or "Which pull requests need my review?" The Azure DevOps MCP Server bridges this gap by connecting your AI assistant directly to your Azure DevOps data.
The Azure DevOps MCP Server provides contextual intelligence based on your actual project data, not generic responses. You can ask natural language questions about your work items, sprints, and releases, and receive insights that understand your team's specific processes and terminology. This process eliminates context switching between tools, provides instant answers without navigating through the Azure DevOps web interface, and automates routine project management tasks through natural language.
Security and privacy
The Azure DevOps MCP Server prioritizes data security and privacy:
- No external API calls: The server doesn't make external API calls that could expose sensitive project information
- User control: You maintain full control over what data your AI assistant can access
- Secure integration: Works seamlessly with your existing AI coding environments without compromising security
- Private data handling: Your sensitive project information never leaves your network while still delivering enterprise-grade AI capabilities
What does MCP Server do?
The Azure DevOps MCP Server enables a two-step process: data retrieval and AI analysis.
1. Data retrieval (MCP Server)
The server provides secure access to your Azure DevOps data:
- Projects and teams: Organization structure and team information
- Work items: Assigned tasks, bugs, user stories, and their details
- Pull requests: Code review status, changes, and linked work items
- Builds and pipelines: CI/CD status, test results, and deployment information
- Test plans: Test cases, results, and coverage data
2. AI analysis (Your AI assistant)
Your AI assistant processes this data to provide:
- Intelligent summaries: Sprint progress, team velocity, and project health
- Actionable insights: Risk identification, bottleneck analysis, and recommendations
- Context-aware responses: Answers that understand your specific projects and processes
Example: You ask "What's blocking our sprint?" → MCP Server retrieves your current work items → Your AI assistant analyzes the data and identifies specific blockers with recommendations.
Common use cases and examples
For more examples, see Example usage.
Tip
To avoid using stale or cached data from previous queries, add to your prompt, "Do not use previously fetched data."
Daily standup preparation
Component
Description
MCP Server provides
Access to your assigned work items and recent updates
AI assistant action
Analyzes the data to help you prepare for standup
Example prompt
"Get my work items for project MyProject and help me prepare for standup - what did I complete, what am I working on, and what's blocked?"
Query translation
WIQL query to retrieve your assigned work items, filtered by project and excluding removed items, ordered by most recent changes
Raw data result
Six work items including bugs, tasks, user stories, and features with their current states (Resolved, Done, Active, New, Blocked)
AI assistant output
| Status | Work Items |
|---|---|
| ✅ Completed | Bug #4521: "Sign in page crashes on mobile Safari" - Resolved yesterdayTask #4498: "Update API documentation for v2.1" - Completed 2 days ago |
| 🔄 In Progress | User Story #4456: "Implement two-factor authentication" - In Progress (three days)└── Subtasks: Database schema (Done), UI components (In Progress), Testing (Not Started)Bug #4534: "Performance issue with large datasets" - Active (one day) |
| ⚠️ Blocked | Feature #4423: "Integration with external payment gateway" - Blocked└── Waiting for: Security team approval (blocked five days)└── Action needed: Follow up with Sarah from InfoSec teamTask #4467: "Deploy to staging environment" - Impediment└── Issue: Staging server maintenance scheduled today 2-4 PM |
| Summary for standup | - Yesterday: Resolved mobile Safari sign-in issue, completed API docs.- Today: Continue 2FA implementation, investigate dataset performance issue.- Blockers: Need security approval for payment integration, staging deployment delayed due to maintenance. |
Sprint planning support
Component
Description
MCP Server provides
Access to backlog items, team data, and work item details
AI assistant action
Helps analyze capacity and plan the sprint
Example prompt
"Get all work items in the product backlog for MyProject and prioritize them for a two-week sprint with three developers"
Query translation
WIQL query to retrieve backlog items with story points, priority, and state information, ordered by priority and stack rank
Raw data result
Nine backlog items including epics, features, user stories, bugs, and tasks with story points and dependencies, plus team metrics (three developers, 14-day sprint, 60-point velocity)
AI assistant output
| Category | Work Items | Points |
|---|---|---|
| ✅ Recommended | Epic #3401: "User Profile Enhancement" - High business value, no dependenciesBug #3456: "Search results time out on large datasets" - Critical user impactFeature #3378: "Export dashboard to PDF" - Frequently requestedStory #3489: "Implement password reset flow" - Security priorityTask #3512: "Upgrade React components to v18" - Technical debtStory #3467: "Add dark mode toggle" - Quick win | 60 total(13 + 8 + 13 + 8 + 13 + 5) |
| ⏳ Deferred | Epic #3234: "Mobile app rewrite" - Too large for single sprintFeature #3445: "AI-powered recommendations" - Waiting for data science teamStory #3523: "Integration with external CRM" - Blocked by vendor API | 34, 21, 13 |
| Planning recommendations | - Balance: Mix of features (3), bugs (1), technical debt (1), quick wins (1)- Risk level: Low-medium risk items for predictable delivery- Dependencies: All selected items have resolved dependencies- Skills: Work distribution matches current team expertise- Business value: Prioritized customer-facing features and critical fixes |
Code review workflow
Component
Description
MCP Server provides
Pull request details and linked work item information
AI assistant action
Provides business context for technical changes
Example prompt
"Get details for pull request #67 and its linked work items and help me understand the business effect of these code changes"
API translation
REST API calls to retrieve pull request details and linked work item information with expanded relations
Raw data result
Pull request #67 (caching layer implementation, eight files changed, +156/-42 lines) and two linked work items (performance user story and settings bug)
AI assistant output
| Component | Details |
|---|---|
| Pull Request | PR #67: "Implement caching layer for user preferences"Author: Christie Church, Status: Active (awaiting review)Files: 8 changed (+156, -42 lines), Created: Two days ago |
| Linked Work Items | 🎯 User Story #2847: "Improve app performance for returning users" (High priority)└── Target: Page load time < 2 seconds for logged-in users🐛 Bug #2901: "Settings page loads slowly for users with many preferences" (Medium severity)└── Impact: Affects 15% of premium users, causing support tickets |
| Business impact summary | - 💼 Strategic Alignment: Supports Q3 user retention goal (20% improvement)Addresses premium user experience issues affecting revenueEnables future personalization features- 📊 Expected Outcomes: Performance: 60% reduction in preference loading time (2.5s → 1.0s)User Experience: Eliminates loading delays in settings and profile pagesSupport Impact: Expected 40% reduction in performance-related tickets- 🔍 Review Focus Areas: Cache invalidation logic (data consistency)Error handling for cache unavailabilityPerformance monitoring implementationSecurity considerations for cached user data |