Amami MCP docs
Quickstart
Use this path when you want an AI assistant to help you register or log in, authorize Amami MCP, connect a website, and start analyzing growth. If you are still choosing the deployment model, read Installation options first. If you are new to the product model, read Core analytics first.
Requirements
- Node.js 20 or newer.
- npm or another Node package runner.
- A browser on your machine for Amami login, registration, and authorization.
- An MCP-capable client such as Codex, Claude, Cursor, Windsurf, Cline, Continue, Zed, VS Code, OpenCode, or a custom script.
Check your local runtime:
node --version
npm --version
One prompt setup
Copy this prompt into your AI assistant:
Install the Amami MCP server, then guide me through browser login and authorization.
https://dashboard.amami.dev/install/mcp-install.md
The assistant should install amami-analytics-mcp, run browser setup, help you configure the MCP client, and verify that the tools work.
Browser setup
You can also run setup yourself:
npx -y amami-analytics-mcp setup --write
Expected flow:
- Amami opens in your browser.
- You log in or register.
- You click Authorize MCP.
- The CLI creates a local API key.
- Credentials are saved to
~/.amami-analytics-mcp/.env. - The MCP server auto-loads that file on future runs.
Use read-only setup if you only want analytics inspection:
npx -y amami-analytics-mcp setup
Configure the MCP client
After browser setup, most clients can use:
{
"mcpServers": {
"amami": {
"command": "npx",
"args": ["-y", "amami-analytics-mcp@latest"]
}
}
}
Client-specific setup pages:
Connect a website
After the MCP client loads, ask:
Create a tracked website for this project, give me the tracking script, and tell me exactly where to add it.
The tracking script uses this shape:
<script defer src="https://dashboard.amami.dev/script.js" data-website-id="WEBSITE_ID"></script>
For a deeper explanation of pageviews, framework snippets, events, UTM parameters, and conversion tracking, read Tracking and events. For goals and multi-step conversion paths, read Goals and funnels.
First events to add
Do not wait until the dashboard looks empty. Add a few meaningful events early:
- Signup CTA clicked.
- Install command copied.
- MCP config copied.
- Website connected.
- First event received.
- Pricing CTA clicked.
These events make later AI analysis much more useful because the assistant can reason about intent, not just traffic volume.
Verify tracking
Ask:
Verify that Amami can see this website, send a test event if write tools are enabled, and summarize the first metrics I should monitor.
Useful verification tools include get_me, list_websites, create_website, send_event, get_stats, get_pageviews, get_metrics, and get_active_visitors.
Ask your first analytics question
After verification, ask a specific question:
Analyze the last 7 days for this website. Compare visitors, top pages, referrers, and conversion events. Tell me what changed and what I should improve first.
For more prompt patterns, read Natural language queries.