Matomo MCP Server
Enables LLMs to directly query a Matomo analytics instance, execute reporting methods, fetch historical trends, report metadata, and dynamically generated chart images through typed MCP tools.
README
Matomo MCP Server
Matomo Direct MCP (Model Context Protocol) Server is a robust, lightweight TypeScript implementation that exposes the full power of the Matomo Analytics HTTP Reporting API directly to LLM agents.
Unlike intermediary solutions, this server connects directly to your Matomo instance, allowing clients (like Claude Desktop, Cursor, or custom MCP wrappers) to run reporting queries, fetch historical trend analysis, inspect report metadata, and retrieve dynamically generated graph images—all through beautifully typed, auto-documented MCP tools.
🚀 Key Features
- Direct Integration – No proxy databases or external authentication gateways; queries the Matomo HTTP API directly using your
token_auth. - Full Reporting Coverage – Call any API method from any core or custom plugin.
- Auto-Validated Inputs – Every tool payload is checked against strict
JSON-Schemaschemas via the MCP SDK and Zod. - Image Generation Support – Directly outputs binary PNG charts (
ImageGraph.get) for visualization-capable LLM interfaces. - Resilience & Reliability – Configurable request timeouts and robust HTTP retry logic with exponential back-off.
- Flexible Environments – Fully supports command-line flags, environment variables, and Docker deployment.
🛠️ Technology Stack
- Language: TypeScript (ESNext Modules, Strict Mode)
- Runtime: Node.js (Recommended
>= 22.0.0for native fetch) - Server Framework: @modelcontextprotocol/sdk (v1.15.0)
- Configuration & Validation: Zod (v3.23.8)
- Development Tooling: tsx (watch mode/TS execution), typescript compiler (v5.6.x)
📐 Project Architecture
The Matomo Direct MCP Server operates as a STDIO transport process. It communicates via system standard input/output streams and is fully stateless.
┌─────────────────┐ STDIO ┌─────────────────────────┐
│ │ ◄───────────────────────────► │ │
│ MCP Client │ (List Tools, Call Tool) │ Matomo Direct MCP Srv │
│ (Claude/Cursor) │ ────────────────────────────► │ (TypeScript Process) │
└─────────────────┘ └───────────┬─────────────┘
│
HTTPS Fetch API
(token_auth, JSON)
│
▼
┌─────────────────────────┐
│ Matomo Analytics │
│ HTTP Reporting API │
└─────────────────────────┘
When started, the server:
- Parses & Validates Configuration: Parses CLI flags or environment variables to find the
MATOMO_HOSTandMATOMO_TOKEN_AUTH. - Verifies Connectivity: Performs a non-blocking startup check against
API.getMatomoVersionto ensure the host is reachable. - Exposes Typed Tools: Publishes structured JSON-schemas describing all report actions to the client.
- Handles Requests: Receives tool invocations, dynamically builds requests (serializing parameters and authentication tokens), executes HTTP queries, detects content-types (JSON, XML, TSV, CSV, PNG), and formats clean, structured outputs back to the agent.
📦 Project Structure
mcp-matomo/
├── src/
│ ├── index.ts # Standalone CLI entrypoint, configuration parsing, and startup
│ └── server.ts # Core MCP server definition, API client, and tool handlers
├── dist/ # Compiled production JavaScript files (generated via tsc)
├── package.json # Dependencies, compilation scripts, and metadata
├── tsconfig.json # Strict TypeScript compiler options
├── .gitignore # Comprehensive secrets & artifact ignore rules
└── README.md # Documentation (this file)
🚥 Getting Started
Prerequisites
- Node.js
>= 22.0.0 - npm
>= 10.0.0 - A running Matomo Analytics instance with an API token (
token_auth).
Installation
-
Clone the repository to your local machine:
git clone https://github.com/alexgenovese/matomo-mcp.git cd matomo-mcp -
Install development and runtime dependencies:
npm ci
⚙️ Configuration
The server can be configured seamlessly using either Command Line Arguments or Environment Variables. CLI flags take priority over environment variables.
| CLI Argument | Environment Variable | Default | Description |
|---|---|---|---|
--matomo-host=URL |
MATOMO_HOST |
Required | Base URL of your Matomo instance (e.g., https://analytics.yourcompany.com/). |
--matomo-token=TOKEN |
MATOMO_TOKEN_AUTH |
Required | Your Matomo secret API token (token_auth). |
--timeout=MS |
REQUEST_TIMEOUT |
30000 |
HTTP request timeout in milliseconds. |
--retry=COUNT |
RETRY_COUNT |
3 |
Maximum HTTP request retries. |
--retry-delay=MS |
RETRY_DELAY |
1000 |
Initial retry delay in ms. Multiplied exponentially on consecutive failures. |
--format=FORMAT |
MATOMO_DEFAULT_FORMAT |
json |
Default response format for reporting queries (e.g., json, xml). |
🏃 Running the Server
1. Local Development (with Hot Reloading)
Run the server directly from TypeScript source code using tsx:
npm run dev -- --matomo-host=https://your-matomo-url.com --matomo-token=your_token_auth
2. Production Build & Execution
Compile the TypeScript project to production-ready ES modules, then execute using Node:
# Compile TS to JS in /dist
npm run build
# Run stand-alone server
node dist/index.js --matomo-host=https://your-matomo-url.com --matomo-token=your_token_auth
3. Running with Docker
You can containerize and run the server inside a Docker environment:
# Build the Docker image
docker build -t matomo-mcp .
# Run the container
docker run -i --rm -e MATOMO_HOST="https://your-matomo-url.com" -e MATOMO_TOKEN_AUTH="your_token_auth" matomo-mcp
🛠️ MCP Tools Overview
The server exposes six highly focused, dynamically parameterized tools:
1. matomo_call
Generic, escape-hatch endpoint to run any Matomo HTTP API reporting method.
- Required Arguments:
method(string): TheModule.actionsignature (e.g.,VisitsSummary.getorUsersManager.getUsers).
- Optional Arguments:
params(object): Key-value pairs representing request arguments.format(enum): Output serialization format (json,xml,csv,tsv,html,rss,original,png).
2. matomo_list_report_metadata
Fetch the complete metadata suite of compatible reports for a site.
- Required Arguments:
idSite(integer): ID of the Matomo website.
- Optional Arguments:
period,date,language,showSubtableReports,hideMetricsDoc.
3. matomo_get_metadata
Query granular metadata schemas, documentation, and specific column dimensions of a single report action.
- Required Arguments:
idSite(integer),apiModule(string),apiAction(string).
4. matomo_get_processed_report
Retrieve high-level processed metrics, calculations, and tables complete with arithmetic.
- Required Arguments:
idSite(integer),apiModule(string),apiAction(string).
- Optional Arguments:
segment(segment filter string),flat(boolean),expanded(boolean),filter_limit(integer).
5. matomo_get_row_evolution
Analyze historical trend lines for a specific report row label over time.
- Required Arguments:
idSite(integer),apiModule(string),apiAction(string),label(string).
6. matomo_get_image_graph
Acquire raw PNG chart visualizations directly generated by Matomo's Graph Engine.
- Required Arguments:
idSite(integer),apiModule(string),apiAction(string).
- Optional Arguments:
graphType(evolution, horizontalBar, verticalBar, pie),width,height,columns.
🔌 Integration Guides
Claude Desktop Integration
To register this server with your local Claude Desktop app, edit your configuration file (usually located at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or %APPDATA%\Claude\claude_desktop_config.json on Windows) and add the following entry:
{
"mcpServers": {
"matomo-direct-mcp": {
"command": "node",
"args": [
"/Users/alexgenovese/Documents/GitHub/mcp/mcp-matomo/dist/index.js",
"--matomo-host=https://your-matomo-instance.com",
"--matomo-token=YOUR_MATOMO_TOKEN_AUTH"
]
}
}
}
Note: Ensure you have compiled the server beforehand by running npm run build.
💻 Development Workflow
Coding Standards
We enforce highly clean, structured, and modern development standards:
- Strict Types: Explicit typings everywhere;
anyis restricted only to generic API parsing. - Modular ES Imports: Node.js ES Modules are strictly enforced. Relative file imports must include the
.jsextension (e.g.import { Server } from "./server.js"). - Robust Error Handling: Network failures are monitored and retried using structured back-offs, while standard error objects are printed to standard error (
console.error) so as not to corrupt the STDIO transport on standard output.
Linting
Validate codebase health and alignment to guidelines:
npm run lint
🤝 Contributing
Contributions are welcome! Please follow these simple steps to contribute:
- Fork the repository and create your feature branch (
git checkout -b feature/amazing-feature). - Verify changes build perfectly and conform to standard formatting guidelines.
- Open a Pull Request detailing the purpose and scope of your modifications.
📄 License
This project is licensed under the MIT License - see the package.json file for licensing declarations.
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。