Amami

Amami is AI-native web analytics built for agents, not dashboards. Connect your site, then ask Claude Code, Cursor, or Codex about traffic, referrers, and conversions. The MCP server returns real numbers (visits, bounce rate, top sources) in natural language, with citations.

The product runs on a trusted analytics foundation (likely Umami-compatible) and adds browser-authorized MCP access. You bring your own keys; the site never sees your API credentials. The free tier includes 100K events/month across five websites. Write tools (creating sites, sending events) require explicit opt-in.

The live dashboard and documentation are polished. The founder's indie-hackers post describes three months of dogfooding and the shift from dashboards to agent-native analytics. This is a working example of how MCP can turn a traditional SaaS product into an AI-first workflow.