An MCP server for on-prem website analytics
Flowpoint includes a built-in MCP server so the AI clients your team already uses can read your analytics and build dashboards — without your data ever leaving your environment. Connect Claude or any MCP client, point it at your own model, and let your agent do the analysis.
What the MCP server does
Query analytics as MCP tools
Traffic, sessions, events, sources, and funnels are exposed as structured tools your agent can call directly — no SQL, no scraping the UI.
Build dashboards from your client
Ask your agent to assemble charts, tables, and reports. Flowpoint builds the dashboard interface while your data is read locally.
Bring your own model
Run inference on a local model inside your network, or connect a cloud model like Claude. The MCP server never ships your raw analytics to the model.
On-prem and private
The server runs as part of your Flowpoint deployment. Every request stays inside your infrastructure — nothing is proxied through us.
Connect your client
Add the Flowpoint MCP server to your client's configuration and it appears as a set of tools your agent can call. Because everything runs on-prem, you control the endpoint, the credentials, and the model.
- Claude & Claude Code
- Cursor
- Any MCP-compatible client
// Example MCP client config
{
"mcpServers": {
"flowpoint": {
"url": "https://analytics.your-company.internal/mcp",
"headers": {
"Authorization": "Bearer <your-token>"
}
}
}
}Endpoint and auth are configured in your own deployment. See the docs for setup.
Deploy Flowpoint in your environment
Run agentic website analytics on your own infrastructure — your data, your model, fully private. Tell us about your setup and we'll get you licensed and deployed.
On-premise · Your data never leaves your infrastructure