Umami MCP Server

Umami MCP Server

Read-only MCP server for Umami analytics. It talks to the Umami REST API directly over HTTP, supporting self-hosted and cloud setups.

Category
访问服务器

README

Umami MCP Server

Read-only MCP server for Umami analytics. It talks to the Umami REST API directly over HTTP, supports self-hosted Umami first, and also works with Umami Cloud API keys.

This repo is built so upper-layer agents can ask for analytics without re-reading the Umami docs each time. No browser automation, no DOM scraping, no write operations.

Features

  • Read-only MCP tools for the most common Umami analytics queries
  • Two auth modes:
    • UMAMI_API_KEY
    • UMAMI_USERNAME + UMAMI_PASSWORD
  • Self-hosted bearer token caching in memory
  • Automatic re-login and one retry on 401 for username/password mode
  • ISO time strings and millisecond timestamps both accepted
  • Shared filter handling across stats, pageviews, breakdowns, and event series
  • Strict TypeScript with small, maintainable modules
  • Clear structured tool output and explicit error categories

Requirements

  • Node.js >= 20
  • pnpm >= 10

Install

From npm:

npm install -g umami-analytics-mcp

Or run without installing:

npx -y umami-analytics-mcp

For local development in this repo:

pnpm install

Configure

Create a .env file from .env.example.

cp .env.example .env

Fill in one of the following auth options:

  1. API key mode
UMAMI_API_URL=https://api.umami.is/v1
UMAMI_API_KEY=your-api-key
UMAMI_DEFAULT_TIMEZONE=Asia/Shanghai
  1. Self-hosted username/password mode
UMAMI_API_URL=https://umami.example.com/api
UMAMI_USERNAME=admin
UMAMI_PASSWORD=secret
UMAMI_DEFAULT_TIMEZONE=Asia/Shanghai

Notes:

  • UMAMI_API_URL should point to the API base, not just the site origin.
  • If both UMAMI_API_KEY and username/password are set, UMAMI_API_KEY wins.
  • UMAMI_DEFAULT_TIMEZONE is used when a tool supports timezone and you omit it.

Run Locally

Development mode:

pnpm dev

Build and run:

pnpm build
pnpm start

The server uses MCP stdio transport, so it stays attached to stdin/stdout until the client disconnects.

If installed from npm, the equivalent command is:

umami-analytics-mcp

Test

pnpm test
pnpm build

There is also a real-API integration test template that is skipped by default:

UMAMI_INTEGRATION_TEST=1 pnpm test:integration

MCP Inspector

Build first:

pnpm build

Then launch the official MCP Inspector against the built server:

npx @modelcontextprotocol/inspector node dist/cli.js

For hot reload during development, this also works:

npx @modelcontextprotocol/inspector pnpm dev

If you want to test the published package path instead, use:

npx @modelcontextprotocol/inspector npx -y umami-analytics-mcp

Make sure the same Umami environment variables are available to the Inspector process.

Recommended smoke calls in Inspector:

  1. umami_ping
  2. umami_list_websites
  3. umami_get_stats
  4. umami_get_breakdown

Tools

umami_ping

Validates configuration and authentication.

Example:

{}

umami_list_websites

Lists accessible websites.

Example:

{}

umami_find_website

Fuzzy search by website name or domain.

Example:

{
  "query": "example.com"
}

umami_get_stats

Summary stats for a website and time range.

Example:

{
  "websiteId": "8f2f8ce2-1234-4567-89ab-0123456789ab",
  "startAt": "2026-04-23T00:00:00+08:00",
  "endAt": "2026-04-23T23:59:59+08:00",
  "filters": {
    "path": "/pricing"
  }
}

umami_get_pageviews

Time-series pageviews and sessions.

Example:

{
  "websiteId": "8f2f8ce2-1234-4567-89ab-0123456789ab",
  "startAt": "2026-04-17T00:00:00+08:00",
  "endAt": "2026-04-23T23:59:59+08:00",
  "unit": "day",
  "compare": "prev",
  "filters": {
    "path": "/blog"
  }
}

umami_get_breakdown

Breakdown rows such as top pages, referrers, countries, browsers, devices, and more.

Example:

{
  "websiteId": "8f2f8ce2-1234-4567-89ab-0123456789ab",
  "startAt": "2026-04-17T00:00:00+08:00",
  "endAt": "2026-04-23T23:59:59+08:00",
  "type": "path",
  "limit": 10,
  "expanded": false
}

umami_get_active

Returns the current active visitor count.

Example:

{
  "websiteId": "8f2f8ce2-1234-4567-89ab-0123456789ab"
}

umami_get_events_series

Returns custom event counts over time.

Example:

{
  "websiteId": "8f2f8ce2-1234-4567-89ab-0123456789ab",
  "startAt": "2026-04-17T00:00:00+08:00",
  "endAt": "2026-04-23T23:59:59+08:00",
  "unit": "day"
}

Shared Filters

These tools support the same filters object:

  • path
  • referrer
  • title
  • query
  • browser
  • os
  • device
  • country
  • region
  • city
  • hostname

Error Handling

Tool errors are returned as structured MCP tool results with these categories:

  • config_missing
  • auth_failed
  • website_not_found
  • umami_http_error
  • network_timeout
  • network_error
  • invalid_input

Mount In Any Stdio MCP Client

Any stdio-based MCP client can run this server with:

  • command: npx
  • args: ["-y", "umami-analytics-mcp"]
  • env: your Umami variables

Example JSON snippet for clients that use an mcpServers object:

{
  "mcpServers": {
    "umami": {
      "command": "npx",
      "args": ["-y", "umami-analytics-mcp"],
      "env": {
        "UMAMI_API_URL": "https://umami.example.com/api",
        "UMAMI_USERNAME": "admin",
        "UMAMI_PASSWORD": "secret",
        "UMAMI_DEFAULT_TIMEZONE": "Asia/Shanghai"
      }
    }
  }
}

For a local unreleased checkout, you can still point directly to the built file:

{
  "mcpServers": {
    "umami": {
      "command": "node",
      "args": ["/absolute/path/to/umami-mcp/dist/cli.js"]
    }
  }
}

Publish To npm

This repo is now structured as an npm CLI package.

Recommended release flow:

pnpm test
pnpm build
pnpm pack --dry-run
npm login
pnpm publish

Notes:

  • The package name is set to umami-analytics-mcp because umami-mcp is already taken on npm.
  • The current license is UNLICENSED as a safe placeholder. Replace it before a real public open-source release if you want a permissive license.
  • If you later publish under your own npm scope, change only the name field in package.json.

Internal Docs

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选