metabase-mcp-navi
A MCP server for Metabase that gives AI assistants direct access to dashboards, cards, and query execution.
README
Metabase MCP Server
A Model Context Protocol (MCP) server for Metabase - gives AI assistants direct access to dashboards, cards, and query execution.
Features
- Three Authentication Methods: API Key, Session ID, or Username/Password
- Dashboard Tools: List, search, get details, get cards from dashboards
- Card Tools: List, search, get details, execute saved questions
- Database Tools: List databases, tables, get metadata, sync
- Query Tools: Execute SQL, test queries, get suggestions, explain queries
- Image Export: Export cards and dashboards as PNG images
- CRUD Operations: Create, update, delete cards, dashboards, and collections
Quick Start
Installation
# Install via pip
pip install metabase-mcp-navi
# Or via uvx (recommended for MCP)
uvx metabase-mcp-navi
Configuration
Set environment variables:
# Required
export METABASE_URL=https://your-metabase-instance.com
# Authentication (choose one):
export METABASE_API_KEY=your_api_key
# OR
export METABASE_SESSION_ID=your_session_id
# OR
export METABASE_USER_EMAIL=your_email
export METABASE_PASSWORD=your_password
# Optional
export METABASE_VERIFY_SSL=false # Disable SSL verification
export METABASE_TIMEOUT=60 # Request timeout in seconds
Cursor MCP Configuration
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"metabase": {
"command": "uvx",
"args": ["metabase-mcp-navi"],
"env": {
"METABASE_URL": "https://your-metabase-instance.com",
"METABASE_SESSION_ID": "your_session_id",
"METABASE_VERIFY_SSL": "false"
}
}
}
}
Claude Desktop Configuration
Add to your Claude Desktop config:
{
"mcpServers": {
"metabase": {
"command": "uvx",
"args": ["metabase-mcp-navi"],
"env": {
"METABASE_URL": "https://your-metabase-instance.com",
"METABASE_SESSION_ID": "your_session_id"
}
}
}
}
Available Tools
Dashboard Tools
| Tool | Description |
|---|---|
list_dashboards |
List all accessible dashboards |
get_dashboard |
Get dashboard details |
get_dashboard_cards |
Get all cards from a dashboard |
search_dashboards |
Search dashboards by name |
Card Tools
| Tool | Description |
|---|---|
list_cards |
List saved questions |
get_card |
Get card details and SQL query |
execute_card |
Run a saved question |
search_cards |
Search cards by name |
Database Tools
| Tool | Description |
|---|---|
list_databases |
List connected databases |
get_database |
Get database details |
list_tables |
List tables in a database |
get_table_metadata |
Get table columns and types |
sync_database |
Trigger metadata sync |
Query Tools
| Tool | Description |
|---|---|
execute_query |
Execute native SQL |
test_query |
Test query with auto LIMIT |
get_query_suggestions |
Get suggested queries for a table |
explain_query |
Get query execution plan |
Image Tools
| Tool | Description |
|---|---|
get_card_image |
Get card as base64 PNG |
download_card_image |
Save card image to disk |
get_dashboard_cards_as_images |
Check exportable cards |
download_all_dashboard_cards |
Download all dashboard images |
CRUD Tools
| Tool | Description |
|---|---|
create_collection |
Create a new collection |
create_card |
Create a new card with SQL |
create_dashboard |
Create a new dashboard |
add_card_to_dashboard |
Add card to dashboard |
update_card |
Update card properties |
delete_card |
Delete a card |
delete_dashboard |
Delete a dashboard |
Other Tools
| Tool | Description |
|---|---|
test_connection |
Test Metabase connectivity |
list_collections |
List folders/collections |
get_collection_items |
Browse collection contents |
Usage Examples
After configuring, you can ask your AI assistant:
- "List all dashboards in Metabase"
- "Get dashboard 3301 details"
- "Execute card 12345"
- "Run this SQL on database 104: SELECT * FROM users LIMIT 10"
- "Search for dashboards about 'payments'"
- "Create a new card with this query..."
- "Download all images from dashboard 42"
Getting Session ID
- Open Metabase in your browser and log in
- Open Developer Tools (F12)
- Go to Application → Cookies
- Copy the value of
metabase.SESSION
Project Structure
metabase-mcp-navi/
├── src/
│ └── metabase_mcp_navi/
│ ├── __init__.py # Package initialization
│ ├── __main__.py # Module entry point
│ ├── config.py # Configuration management
│ ├── client.py # Metabase API client
│ ├── models.py # Data models
│ ├── server.py # MCP server setup
│ └── tools/
│ ├── __init__.py # Tools package
│ ├── dashboards.py # Dashboard tools
│ ├── cards.py # Card/question tools
│ ├── databases.py # Database tools
│ ├── queries.py # Query execution tools
│ ├── images.py # Image export tools
│ └── crud.py # CRUD operations
├── pyproject.toml # Package configuration
├── README.md
└── LICENSE
Development
# Clone the repository
git clone https://github.com/manish-coder-1007/metabase-mcp-navi.git
cd metabase-mcp-navi
# Install in development mode
pip install -e ".[dev]"
# Run tests
pytest
# Format code
black src/
ruff check src/
License
MIT License - see LICENSE for details.
Author
Manish Balot
Related Projects
- trino-mcp-navi - MCP server for Trino databases
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。