arcade-azure-devops-mcp
An MCP server that enables AI assistants to interact with Azure DevOps projects, work items, repositories, pipelines, wikis, and more via 21 tools.
README
arcade-azure-devops-mcp
An MCP server for Azure DevOps built with Arcade.dev. Enables AI assistants to interact with Azure DevOps projects, work items, repositories, pipelines, wikis, and more.
Features
21 tools covering:
- Core: List projects, teams, search identities
- Work Items: Get, create, update, query work items; add comments
- Repositories: List repos, branches, pull requests; create PRs
- Pipelines: List definitions, builds; queue builds; run pipelines
- Wikis: List wikis, get wiki pages
- Search: Search code across repositories
Quick Start
1. Install dependencies
uv sync
2. Configure credentials
You have two options for providing Azure DevOps credentials:
Option A: Environment Variables (.env file)
Copy the example environment file and add your credentials:
cp .env.example .env
Edit .env:
AZURE_DEVOPS_ORG=your-organization-name
AZURE_DEVOPS_PAT=your-personal-access-token
Option B: Arcade Secrets (recommended for production)
If no .env file or environment variables are found, the server automatically falls back to Arcade Secrets. This is ideal for:
- Publishing to the Arcade marketplace
- Sharing the toolkit without exposing credentials
- Production deployments via Arcade Deploy
Configure secrets in Arcade:
# Login to Arcade
arcade login
# Set your secrets
arcade secrets set AZURE_DEVOPS_ORG "your-organization-name"
arcade secrets set AZURE_DEVOPS_PAT "your-personal-access-token"
Or configure via the Arcade Dashboard.
How it works:
- Server checks for environment variables first
- If not found, requests secrets from Arcade Cloud via
context.get_secret() - Arcade prompts user to authorize secret access (first time only)
3. Create a Personal Access Token (PAT)
- Go to
https://dev.azure.com/{your-org}/_usersSettings/tokens - Click "New Token"
- Select scopes: Code (Read/Write), Work Items (Read/Write), Build (Read/Execute), Wiki (Read/Write)
- Copy the token
4. Run the MCP Server
# stdio transport (default) - for Claude Desktop, CLI tools
uv run server.py stdio
# http transport - for Cursor, VS Code
uv run server.py http
For HTTP transport, view the API docs at http://127.0.0.1:8000/docs
Configure MCP Clients
Cursor IDE
arcade configure cursor
Or manually add to .cursor/mcp.json:
With environment variables:
{
"mcpServers": {
"azure-devops": {
"command": "uv",
"args": ["run", "server.py", "stdio"],
"cwd": "/path/to/arcade-azure-devops-mcp",
"env": {
"AZURE_DEVOPS_ORG": "your-org",
"AZURE_DEVOPS_PAT": "your-pat"
}
}
}
}
With Arcade Secrets (no local credentials):
{
"mcpServers": {
"azure-devops": {
"command": "uv",
"args": ["run", "server.py", "stdio"],
"cwd": "/path/to/arcade-azure-devops-mcp"
}
}
}
VS Code
arcade configure vscode
Claude Desktop
arcade configure claude
Available Tools
| Category | Tool | Description |
|---|---|---|
| Core | list_projects |
List all projects in the organization |
get_project |
Get project details | |
list_teams |
List teams in a project | |
search_identities |
Search for users/groups | |
| Work Items | get_work_item |
Get a work item by ID |
create_work_item |
Create a new work item | |
update_work_item |
Update an existing work item | |
run_work_item_query |
Run a WIQL query | |
my_work_items |
Get work items assigned to you | |
add_work_item_comment |
Add a comment to a work item | |
| Repos | list_repositories |
List Git repositories |
list_branches |
List branches in a repository | |
list_pull_requests |
List pull requests | |
create_pull_request |
Create a new PR | |
| Pipelines | list_build_definitions |
List pipeline definitions |
list_builds |
List builds | |
queue_build |
Queue a new build | |
run_pipeline |
Start a pipeline run | |
| Wikis | list_wikis |
List wikis |
get_wiki_page |
Get wiki page content | |
| Search | search_code |
Search code across repositories |
Architecture
Request Flow
┌────────────┐ ┌────────────┐ ┌──────────────┐ ┌─────────────┐
│ MCP Client │ │ MCP Server │ │ Arcade Cloud │ │ Azure DevOps│
└─────┬──────┘ └─────┬──────┘ └──────┬───────┘ └──────┬──────┘
│ │ │ │
│ Call tool │ │ │
│──────────────────>│ │ │
│ │ │ │
│ │ Check env vars │ │
│ │◄──────────────────►│ │
│ │ │ │
│ │ [If env vars missing - paid tier] │
│ │ get_secret() │ │
│ │───────────────────>│ │
│ │<───────────────────│ │
│ │ │ │
│ │ API Request (Basic Auth) │
│ │─────────────────────────────────────────>│
│ │<─────────────────────────────────────────│
│ │ │ │
│ Tool result │ │ │
│<──────────────────│ │ │
│ │ │ │
<details> <summary>View Mermaid diagram (for GitHub/GitLab)</summary>
sequenceDiagram
participant C as MCP Client
participant S as MCP Server
participant A as Arcade Cloud
participant D as Azure DevOps
C->>S: Call tool (e.g., list_projects)
Note over S: Resolve Credentials
S->>S: Check env: AZURE_DEVOPS_ORG
S->>S: Check env: AZURE_DEVOPS_PAT
alt Environment variables exist
S->>S: Use local credentials
else Environment variables missing (paid tier)
S->>A: context.get_secret("AZURE_DEVOPS_ORG")
A-->>S: Return organization name
S->>A: context.get_secret("AZURE_DEVOPS_PAT")
A-->>S: Return PAT token
end
Note over S,D: API Request
S->>D: GET /{org}/_apis/projects?api-version=7.1
D-->>S: JSON response
S-->>C: Tool result
</details>
Component Overview
┌─────────────────────────────────────────────────────────────────────────────┐
│ MCP CLIENT │
│ (Cursor / Claude Desktop) │
└─────────────────────────────────┬───────────────────────────────────────────┘
│ MCP Protocol (stdio/http)
▼
┌─────────────────────────────────────────────────────────────────────────────┐
│ MCP SERVER │
│ ┌─────────────────┐ ┌──────────────────┐ ┌─────────────────────────┐ │
│ │ MCPApp │ │ Azure DevOps │ │ AuthManager │ │
│ │ Framework │──│ Tools (21) │──│ (credential resolver) │ │
│ └─────────────────┘ └──────────────────┘ └───────────┬─────────────┘ │
└─────────────────────────────────────────────────────────┼───────────────────┘
│
┌─────────────────────────────────┼─────────────────┐
│ │ │
▼ [Priority 1] ▼ [Priority 2] │
┌─────────────────────────┐ ┌───────────────────────┐ │
│ Local .env / │ │ Arcade Cloud │ │
│ Environment Variables │ │ (Secrets API) │ │
└─────────────────────────┘ │ [Paid tier only] │ │
└───────────────────────┘ │
│
┌─────────────────┘
▼
┌───────────────────────┐
│ Azure DevOps │
│ REST API v7.1 │
└───────────────────────┘
<details> <summary>View Mermaid diagram (for GitHub/GitLab)</summary>
flowchart TB
subgraph MCPClient [MCP Client]
Cursor[Cursor / Claude Desktop]
end
subgraph MCPServer [MCP Server]
App[MCPApp Framework]
Tools[Azure DevOps Tools]
Auth[AuthManager]
end
subgraph External [External Services]
Arcade[Arcade Cloud - Secrets API]
ADO[Azure DevOps REST API v7.1]
end
subgraph Local [Local Config]
Env[.env file / Environment Variables]
end
Cursor -->|MCP Protocol| App
App --> Tools
Tools --> Auth
Auth -->|Priority 1| Env
Auth -.->|Priority 2 - Paid Tier| Arcade
Auth -->|Basic Auth| ADO
</details>
Credential Resolution Priority
- Environment variables (
.envfile or system env) - checked first - Arcade Cloud secrets - fallback if env vars not found (requires paid Arcade tier)
Project Structure
arcade-azure-devops-mcp/
├── server.py # MCP server entry point with all tools
├── arcade_azure_devops_mcp/ # Azure DevOps client library
│ ├── __init__.py
│ ├── client.py # REST API client (httpx)
│ ├── models.py # Pydantic models
│ └── auth/ # Authentication
│ ├── __init__.py
│ ├── manager.py # Credential management (env + Arcade secrets)
│ └── oauth.py # OAuth/Azure AD (optional)
├── pyproject.toml # Dependencies & entry points
├── .env.example # Environment template
└── README.md
Development
# Install with dev dependencies
uv sync --all-extras
# Run tests
uv run pytest
Publishing to Arcade
To publish this toolkit to the Arcade marketplace:
# Ensure pyproject.toml has correct entry points
arcade publish
Users can then install via:
arcade install arcade-azure-devops-mcp
License
MIT
Links
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。