PowerBI MCP Server

平台与服务

by gurvinder-dhillon

PowerBI REST API integration. Query workspaces, datasets, and execute DAX queries via MCP.

什么是 PowerBI MCP Server

PowerBI REST API integration. Query workspaces, datasets, and execute DAX queries via MCP.

README

PowerBI MCP Server

PyPI Python Versions License: MIT

<!-- mcp-name: io.github.gurvinder-dhillon/powerbi-mcp -->

A Model Context Protocol (MCP) server that provides tools for interacting with PowerBI REST APIs. This server enables AI assistants like Claude to query PowerBI workspaces, datasets, and execute DAX queries.

[!WARNING] Security Best Practices

  • Never commit credentials to version control
  • Store credentials in .env files (add to .gitignore)
  • Rotate client secrets regularly in Azure AD
  • Use least-privilege access (only grant necessary workspace permissions)
  • This server has read/write access to PowerBI datasets - use with caution

✨ Features

  • Query Your Data: Run DAX queries to extract insights and analyze your PowerBI data directly through conversation
  • Discover Workspaces & Datasets: Explore what data is available across your organization's PowerBI environment
  • Understand Data Models: Get detailed schema information to know what tables, columns, and relationships exist
  • Natural Language to Insights: Ask questions about your data and get answers without opening PowerBI

See all available tools below.

💡 What Can You Do?

ScenarioExample Prompt
Explore available data"What workspaces do I have access to?"
Discover reports"What reports are available in my workspace?"
Understand data schema"Show me the schema for dataset [dataset-name]"
Monitor data freshes"When was this dataset last refreshed?"
Check parameters"What parameters does this dataset accept?"
Query data with DAX"Run a DAX query to get top 10 sales by region from [dataset]"
Analyze data quality"What tables are in the Sales dataset?"
Extract insights"Get the list of all measures in the Financial dataset"

📋 Prerequisites

  • Azure AD Service Principal: Required for authentication. Follow the Azure AD Configuration steps below to set this up.

🔐 Azure AD Configuration

Before installing the server, you need to set up an Azure AD application with PowerBI access.

Create Azure AD App Registration

  1. Go to Azure Portal
  2. Navigate to Azure Active Directory > App registrations
  3. Click New registration
  4. Enter a name (e.g., "PowerBI MCP Server")
  5. Click Register

Get Credentials

After registration, collect these values:

  • Tenant ID: Found in app Overview page (Directory ID)
  • Client ID: Found in app Overview page (Application ID)
  • Client Secret:
    • Go to Certificates & secrets
    • Click New client secret
    • Add description and set expiry
    • Copy the secret Value (you can only see this once!)

Enable Service Principal in PowerBI

  1. Go to PowerBI Admin Portal
  2. Navigate to Tenant settings > Developer settings
  3. Enable Service principals can use PowerBI APIs
  4. Add your app to the security group or enable for entire organization
  5. Click Apply

[!NOTE] Service principals can access workspaces where they've been granted explicit permissions (Admin, Member, or Contributor roles).

Grant Workspace Access

For each workspace you want to access:

  1. Go to the workspace in PowerBI
  2. Click workspace settings (⚙️) > Access
  3. Click Add people or groups
  4. Search for your app name
  5. Assign role: Admin, Member, or Contributor
  6. Click Add

📦 Installation

Method 0: From PyPI (Recommended)

Once published to PyPI, this is the simplest installation method.

Add the following to your MCP client configuration file:

For Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

json
{
  "mcpServers": {
    "powerbi-mcp": {
      "command": "uvx",
      "args": ["powerbi-mcp"],
      "env": {
        "POWERBI_TENANT_ID": "your-tenant-id-here",
        "POWERBI_CLIENT_ID": "your-client-id-here",
        "POWERBI_CLIENT_SECRET": "your-client-secret-here"
      }
    }
  }
}

For Claude Code (./.mcp.json in your project directory):

json
{
  "mcpServers": {
    "powerbi-mcp": {
      "command": "uvx",
      "args": ["powerbi-mcp"],
      "env": {}
    }
  }
}

When using Claude Code, create a .env file in your project directory (where you run Claude Code from):

env
POWERBI_TENANT_ID=your-tenant-id-here
POWERBI_CLIENT_ID=your-client-id-here
POWERBI_CLIENT_SECRET=your-client-secret-here

[!TIP] The .env file should be in your working directory, not where the server is installed.

For OpenCode (~/.config/opencode/opencode.json):

json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "powerbi-mcp": {
      "type": "local",
      "command": ["uvx", "powerbi-mcp"],
      "enabled": true,
      "env": {
        "POWERBI_TENANT_ID": "your-tenant-id-here",
        "POWERBI_CLIENT_ID": "your-client-id-here",
        "POWERBI_CLIENT_SECRET": "your-client-secret-here"
      }
    }
  }
}

After adding the configuration, restart your MCP client.

Method 1: Direct from GitHub (Development)

This method uses uvx to run the server directly from GitHub without cloning the repository.

Add the following to your MCP client configuration file:

For Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

json
{
  "mcpServers": {
    "powerbi-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/gurvinder-dhillon/powerbi-mcp@main",
        "run-server"
      ],
      "env": {
        "POWERBI_TENANT_ID": "your-tenant-id-here",
        "POWERBI_CLIENT_ID": "your-client-id-here",
        "POWERBI_CLIENT_SECRET": "your-client-secret-here"
      }
    }
  }
}

For Claude Code (./.mcp.json in your project directory):

json
{
  "mcpServers": {
    "powerbi-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/gurvinder-dhillon/powerbi-mcp@main",
        "run-server"
      ],
      "env": {}
    }
  }
}

When using Claude Code, create a .env file in your project directory (where you run Claude Code from):

env
POWERBI_TENANT_ID=your-tenant-id-here
POWERBI_CLIENT_ID=your-client-id-here
POWERBI_CLIENT_SECRET=your-client-secret-here

[!TIP] The .env file should be in your working directory, not where the server is installed.

The server will automatically load environment variables from the .env file in your current working directory.

For OpenCode (~/.config/opencode/opencode.json):

json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "powerbi-mcp": {
      "type": "local",
      "command": [
        "uvx",
        "--from",
        "git+https://github.com/gurvinder-dhillon/powerbi-mcp@main",
        "run-server"
      ],
      "enabled": true,
      "env": {
        "POWERBI_TENANT_ID": "your-tenant-id-here",
        "POWERBI_CLIENT_ID": "your-client-id-here",
        "POWERBI_CLIENT_SECRET": "your-client-secret-here"
      }
    }
  }
}

After adding the configuration, restart your MCP client.

Method 2: Local Clone (Contributors)

For contributors who want to run from a local clone:

  1. Clone the repository:
bash
git clone https://github.com/gurvinder-dhillon/powerbi-mcp.git
cd powerbi-mcp
  1. Install dependencies:
bash
uv sync
  1. Add to your MCP client configuration:

For Claude Desktop:

json
{
  "mcpServers": {
    "powerbi-mcp-local": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/powerbi-mcp",
        "run",
        "run-server"
      ],
      "env": {
        "POWERBI_TENANT_ID": "your-tenant-id-here",
        "POWERBI_CLIENT_ID": "your-client-id-here",
        "POWERBI_CLIENT_SECRET": "your-client-secret-here"
      }
    }
  }
}

For Claude Code (./.mcp.json in your project directory):

json
{
  "mcpServers": {
    "powerbi-mcp-local": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/powerbi-mcp",
        "run",
        "run-server"
      ],
      "env": {}
    }
  }
}

For OpenCode (~/.config/opencode/opencode.json):

json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "powerbi-mcp-local": {
      "type": "local",
      "command": [
        "uv",
        "--directory",
        "/absolute/path/to/powerbi-mcp",
        "run",
        "run-server"
      ],
      "enabled": true,
      "env": {
        "POWERBI_TENANT_ID": "your-tenant-id-here",
        "POWERBI_CLIENT_ID": "your-client-id-here",
        "POWERBI_CLIENT_SECRET": "your-client-secret-here"
      }
    }
  }
}

Replace /absolute/path/to/powerbi-mcp with the actual path to your cloned repository.

When using Claude Code or OpenCode, create a .env file in your project directory with your credentials (see Method 1 Claude Code section above for the format).

🚀 Quick Start

Once installed, try these steps to get started:

  1. Verify Connection: Ask Claude "What PowerBI workspaces do I have access to?"
  2. Explore Data: "Show me the datasets in workspace [workspace-name]"
  3. View Schema: "What tables are in dataset [dataset-name]?"
  4. Run a Query: "Execute this DAX query on [dataset-name]: EVALUATE TOPN(10, Sales)"

🛠️ Available Tools

ToolDescriptionKey Parameters
get_workspacesList accessible PowerBI workspacestop, detail
get_datasetsGet datasets from workspace or "My workspace"workspace_id, detail
get_datasetGet detailed dataset info including schemadataset_id, workspace_id, detail
get_reportsList PowerBI reports in workspaceworkspace_id, format, detail
get_refresh_historyGet dataset refresh history with status/timestampsdataset_id, workspace_id, top, format
get_parametersList dataset parameters and their current valuesdataset_id, workspace_id, format, detail
query_datasetExecute DAX queries against datasetdataset_id, dax_query, workspace_id

Tool Details

1. get_workspaces

List PowerBI workspaces accessible to the service principal.

Parameters:

  • top (optional): Number of workspaces to return (default: 100, max: 5000)
  • detail (optional): Level of detail - "concise", "normal", or "full" (default: "normal")

2. get_datasets

Get list of datasets from a workspace or "My workspace".

Parameters:

  • workspace_id (optional): Workspace ID (omit for "My workspace")
  • detail (optional): Level of detail - "concise", "normal", or "full" (default: "normal")

3. get_dataset

Get detailed information about a specific dataset including schema and tables.

Parameters:

  • dataset_id (required): Dataset ID
  • workspace_id (optional): Workspace ID (omit for "My workspace")
  • detail (optional): Level of detail - "concise", "normal", or "full" (default: "normal")

4. get_reports

List PowerBI reports in a workspace.

Parameters:

  • workspace_id (optional): Workspace ID (omit for "My workspace")
  • format (optional): Response format - "markdown" or "json" (default: "markdown")
  • detail (optional): Level of detail - "concise" or "normal" (default: "concise")

5. get_refresh_history

Get refresh history for a dataset showing recent refresh operations.

Parameters:

  • dataset_id (required): Dataset ID
  • workspace_id (optional): Workspace ID (omit for "My workspace")
  • top (optional): Number of refresh records to return (default: 5, max: 60)
  • format (optional): Response format - "markdown" or "json" (default: "markdown")

6. get_parameters

Get parameters defined in a dataset.

Parameters:

  • dataset_id (required): Dataset ID
  • workspace_id (optional): Workspace ID (omit for "My workspace")
  • format (optional): Response format - "markdown" or "json" (default: "markdown")
  • detail (optional): Level of detail - "concise" or "normal" (default: "normal")

Note: Not supported for datasets with SQL, Oracle, Teradata, SAP HANA DirectQuery connections or datasets modified via XMLA endpoint.

7. query_dataset

Execute DAX queries against a dataset.

Parameters:

  • dataset_id (required): Dataset ID
  • dax_query (required): DAX query (must start with "EVALUATE")
  • workspace_id (optional): Workspace ID (omit for "My workspace")

DAX Query Examples

Basic Table Scan:

dax
EVALUATE
'Sales'

Top N with Sorting:

dax
EVALUATE
TOPN(10, 'Sales', [Amount], DESC)

Filtered Results:

dax
EVALUATE
FILTER('Sales', [Year] = 2024)

Calculated Columns:

dax
EVALUATE
ADDCOLUMNS(
    'Sales',
    "Profit", [Revenue] - [Cost]
)

Aggregated Summary:

dax
EVALUATE
SUMMARIZE(
    'Sales',
    'Product'[Category],
    "Total Sales", SUM('Sales'[Amount])
)

⚠️ Troubleshooting

Authentication Errors

Error: "Authentication failed"

  • Verify your POWERBI_TENANT_ID, POWERBI_CLIENT_ID, and POWERBI_CLIENT_SECRET are correct
  • Check that the client secret hasn't expired in Azure AD
  • Ensure the service principal is enabled in PowerBI Admin Portal

Permission Errors

Error: "403 Forbidden" or "Access denied"

  • Verify the service principal has been granted access to the workspace
  • Check that the workspace role is Admin, Member, or Contributor (Viewer is not sufficient for API access)
  • Confirm "Service principals can use PowerBI APIs" is enabled in PowerBI Admin Portal

Connection Issues

Server not appearing in MCP client

  • Restart your MCP client after adding the configuration
  • Check the configuration file syntax (JSON must be valid)
  • For local clone, verify the absolute path is correct

Tools not working

  • Ensure credentials are configured
  • Check the MCP client logs for detailed error messages
  • For Claude Code, verify .env file is in the current working directory

📚 Resources

💬 Feedback and Support

🤝 Contributing

See DEVELOPER_GUIDE.md for information about developing and contributing to this project.

License

This project is licensed under the MIT License.

常见问题

PowerBI MCP Server 是什么?

PowerBI REST API integration. Query workspaces, datasets, and execute DAX queries via MCP.

相关 Skills

MCP构建

by anthropics

Universal
热门

聚焦高质量 MCP Server 开发,覆盖协议研究、工具设计、错误处理与传输选型,适合用 FastMCP 或 MCP SDK 对接外部 API、封装服务能力。

想让 LLM 稳定调用外部 API,就用 MCP构建:从 Python 到 Node 都有成熟指引,帮你更快做出高质量 MCP 服务器。

平台与服务
未扫描172.7k

Slack动图

by anthropics

Universal
热门

面向Slack的动图制作Skill,内置emoji/消息GIF的尺寸、帧率和色彩约束、校验与优化流程,适合把创意或上传图片快速做成可直接发送的Slack动画。

帮你快速做出适配 Slack 的动图,内置约束规则和校验工具,少踩上传与播放坑,做表情包和演示都更省心。

平台与服务
未扫描172.7k

接口测试套件

by alirezarezvani

Universal
热门

扫描 Next.js、Express、FastAPI、Django REST 的 API 路由,自动生成覆盖鉴权、参数校验、错误码、分页、上传与限流场景的 Vitest 或 Pytest 测试套件。

帮你把API与集成测试自动化跑顺,减少回归漏测;能力全面,尤其适合复杂接口场景的QA团队。

平台与服务
未扫描25.3k

相关 MCP Server

Slack 消息

编辑精选

by Anthropic

热门

Slack 是让 AI 助手直接读写你的 Slack 频道和消息的 MCP 服务器。

这个服务器解决了团队协作中需要 AI 实时获取 Slack 信息的痛点,特别适合开发团队让 Claude 帮忙汇总频道讨论或发送通知。不过,它目前只是参考实现,文档有限,不建议在生产环境直接使用——更适合开发者学习 MCP 如何集成第三方服务。

平台与服务
89.7k

by netdata

热门

io.github.netdata/mcp-server 是让 AI 助手实时监控服务器指标和日志的 MCP 服务器。

这个工具解决了运维人员需要手动检查系统状态的痛点,最适合 DevOps 团队让 Claude 自动分析性能数据。不过,它依赖 NetData 的现有部署,如果你没用过这个监控平台,得先花时间配置。

平台与服务
80.0k

by d4vinci

热门

Scrapling MCP Server 是专为现代网页设计的智能爬虫工具,支持绕过 Cloudflare 等反爬机制。

这个工具解决了爬取动态网页和反爬网站时的头疼问题,特别适合需要批量采集电商价格或新闻数据的开发者。不过,它依赖外部浏览器引擎,资源消耗较大,不适合轻量级任务。

平台与服务
72.9k

评论