> "How was traffic last week?"
"I can read visits, pageviews, bounces, and time on site for the selected period, then summarize what changed."
Analytics for AI-native teams
Amami connects your approved MCP client to Umami, so your team can ask clear questions about traffic, pages, visitors, and referrers without context switching.
What changed in our traffic this week?
Visitors increased, led by organic search. The pricing page received the largest share of attention. Compare it with the previous period to decide what to review.
Ask a direct question, then inspect the underlying Umami metrics.
"I can read visits, pageviews, bounces, and time on site for the selected period, then summarize what changed."
"I can compare the pages receiving attention and keep the underlying metrics in the response for you to inspect."
"I can group referrers and countries from your Umami data, then point out the changes worth investigating."
Authorize in your browser, register the active client, then verify without test data.
Copy this prompt and send it to your AI:
Read and follow the canonical Amami MCP and Skills installation guide at https://analytics.amami.dev/install/mcp-install.md. Use this guide as the only setup source. Do not use web search or another installation guide.The canonical guide installs MCP and Amami Skills, checks Node.js, runs browser setup, registers your active MCP client, and verifies access without creating websites or sending events. Login, registration, and authorization stay in your browser, under your control.
Six focused analytics reads, designed for natural-language questions.
Find the Umami sites your assistant is allowed to analyze.
Pull visitors, pageviews, visits, bounces, and time on site.
Read traffic patterns by day or hour for the period you choose.
See which pages are receiving attention.
Read the current active-visitor count.
Inspect where visitors came from.
Your agent can inspect analytics. You retain control of credentials and authorization.
Login, registration, and MCP authorization happen in your browser. Your agent does not perform them for you.
Your API key is stored by your MCP client on your machine, rather than supplied to an AI conversation.
Setup enables the approved client scope. Verification only checks your account, website list, and available tools without creating test data.
Amami is an AI-native read layer for Umami, not a dashboard or BI replacement.