# Vitrus analytics [Health: Active]

**Category:** 🗄️ Databases  
**Repository:** https://github.com/Vitrus-Dev/vitrus  
**GitHub Stars:** 4  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/vitrus-analytics

## Description
Read-only web analytics — AI traffic, revenue, goals, funnels — with the SQL behind every number.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "vitrus-analytics": {
    "url": "https://app.vitrus.dev/mcp"
  }
}
```

## Documentation

## What the Vitrus analytics MCP server does

The Vitrus analytics MCP server exposes website analytics to MCP-compatible assistants without allowing mutations. It covers common reporting questions such as traffic changes, current visitors, acquisition sources, AI assistant referrals, revenue, goal performance, funnel drop-off, user journeys, retention, performance metrics, and recorded errors.

A distinguishing part of the response format is its evidence trail. Metrics include the SQL statement that produced them, the query parameters, the reporting window, and the resulting value. This lets an assistant cite the query behind a number instead of presenting an unsupported assertion. Vitrus also separates AI assistants that send visitors from crawlers that read site content.

## How it works

The hosted MCP endpoint is available at `https://app.vitrus.dev/mcp`. An MCP client connects over HTTP, then authenticates with OAuth or an API key beginning with `vk_`. Each connection is limited to a selected workspace and can be disconnected from the account's connected-app settings.

Vitrus analytics MCP server tools query the underlying analytics data in read-only mode. The repository describes 15 tools, including `query_stats`, `get_realtime`, `get_ai_traffic`, `get_revenue`, `goal_report`, `analyze_funnel`, `get_journeys`, `get_retention`, `get_web_vitals`, `get_errors`, and `get_digest`. The `query_stats` tool can retrieve metrics across 20 dimensions or by day, with filters.

## Setup and configuration

For the hosted service, add the MCP URL to a compatible client and complete authentication. ChatGPT and Claude can use it as a custom connector; Claude Code can add the HTTP server directly, and Cursor provides an MCP installation link. Gemini CLI can install the repository extension.

The analytics engine can also be self-hosted. The repository requires Bun 1.3 or newer. A basic setup clones the project, installs dependencies, builds the tracker, initializes a local `vitrus.db`, adds a site, and starts the service with `VITRUS_PASSWORD` set. The documented start command serves ingestion and the dashboard on port 3000. A site then adds the generated `v.js` script to its HTML head.

## Tools and capabilities

Use the tools to investigate both high-level performance and specific changes:

- Compare metrics over time or across dimensions with filters.
- Inspect live visitors and incoming events.
- Separate assistant referrals from crawler activity.
- Review revenue by currency and the visit's originating channel.
- Analyze ordered funnels, journeys, retention cohorts, and goals.
- Examine Core Web Vitals and application errors.
- Retrieve an analytics digest with evidence attached.

The repository also includes Claude Code skills for weekly reports, traffic-drop investigation, measurement-plan suggestions, and AI-search audits. These skills are separate from the read-only MCP tools.

## Limitations and notes

The MCP interface is read-only, so it is intended for analysis rather than changing tracking configuration or analytics data. Results are scoped to one workspace. The self-hosted dashboard is described as simpler than the hosted dashboard, so hosted and local feature coverage may differ.

Vitrus analytics MCP server does not replace the site tracking setup: a site must send data through the provided tracker before analytics queries can return meaningful results. The repository describes the product as cookie-free and open source, but the MCP connection still requires authentication for hosted workspace access.

_Full upstream README: https://allmcps.com/mcp/vitrus-analytics/readme_