You can connect Userpilot to Claude using the Model Context Protocol (MCP), and once it’s connected, Claude can check an onboarding flow’s completion rate or pull up an NPS trend without you opening the Userpilot dashboard. Userpilot’s OAuth also just works with a new client, no allow-list wait. This guide covers what Userpilot’s MCP server does, both ways to connect it — directly or through a gateway — and what to check before an AI client reads your product’s usage and feedback data.

What the Userpilot MCP server actually does

Userpilot publishes a remote MCP server at mcp.userpilot.io. It exposes your account’s product usage, onboarding flows, and NPS data as tools Claude can call directly, so prompts like these work against your real data instead of a screenshot you’d otherwise have to describe:

  • “How many users finished the new onboarding checklist this week?”
  • “What’s our NPS trend over the last quarter?”
  • “Which step of the signup flow do people drop off at most?”
  • “Summarize the feedback left on the last in-app survey.”

Claude reads through the same account permissions your Userpilot login already has — it’s querying live flow and survey data, not a static export.

The good news: Userpilot’s OAuth just works

Several servers in gate’s directory — Slack, HubSpot, Intercom, Figma — currently only admit AI clients the vendor has allow-listed in advance, which can mean a connection attempt fails at the authorization step for no obvious reason. Userpilot isn’t one of those. Its OAuth is open to any client that asks, the same as Amplitude’s or PostHog’s, so connecting it is a standard consent-screen flow with nothing to wait on. For the general pattern behind why some servers gate this and others don’t, see the unauthorized client error, explained.

Option 1: Connect Userpilot to Claude directly

Claude supports remote MCP servers natively. The short version of the setup:

  • Open Claude’s connector or MCP settings.
  • Add a remote server pointing at Userpilot’s MCP endpoint.
  • Sign in with your Userpilot account when prompted — standard OAuth, so Claude never sees your Userpilot password, and you can revoke access from Userpilot’s own account settings at any time.

That’s enough if Claude is the only AI client you use with Userpilot. The tradeoff shows up once a second client or a second teammate needs the same access: each one repeats the same OAuth grant, and each grant is its own thing to remember later.

Option 2: Connect it once, through a gateway

If you’re already juggling more than one MCP server or more than one AI client, wiring each pair directly gets old fast — the connection sprawl covered in what an MCP gateway is. The idea is to connect Userpilot (and everything else) to one gateway URL, then point every AI client at that single URL instead of repeating the OAuth step per client.

With gate specifically, connecting Userpilot looks like this: add gate’s gateway URL to Claude once, then pick Userpilot from the Userpilot server page in the directory and sign in with OAuth the same way you would directly. From then on, Userpilot’s tools are available in every client connected to your gate URL — Claude, ChatGPT, or Cursor — with the same per-tool allow, ask, or block rule applied everywhere instead of set up separately in each client. See the full list of servers set up this way on the MCP servers page.

What to check before you connect your onboarding and NPS data

A Userpilot account usually holds more than the one flow you had in mind when you connect it — every onboarding checklist, every survey response, every NPS comment across the product. A few things worth checking:

  • Do survey responses carry free text from real users? NPS and in-app survey comments are often written by actual customers, sometimes naming a specific complaint or person. An AI client reading them surfaces whatever was typed, not a redacted summary.
  • Which workspace is actually in scope? If your Userpilot account spans more than one product or environment, know which one the connection points at before you ask Claude to summarize “our onboarding numbers.”
  • Read, or read and manage? Check the actual tool list rather than assuming it’s read-only — gate’s free MCP security scanner lists every tool a server exposes, so you can see whether anything on there can edit a live flow or survey rather than just query one.
  • Who else on the team connects it? If more than one person wires up Userpilot to their own client, a shared, revocable setup beats everyone holding their own grant to the same account — the same question covered in MCP access control for teams.

Troubleshooting the connection

  • Claude can’t find a flow you know exists. Onboarding flows and surveys don’t carry over between separate Userpilot workspaces — check the connection is pointed at the right one.
  • Numbers look different from the dashboard. A metric’s result can depend on the date range and segment it was built with; double-check Claude is reading the same saved view rather than recomputing a similar-sounding one from raw events.
  • Tools disappear after a Userpilot update. Remote servers can change their tool list without warning. If you’re connected through a gateway, that’s exactly what drift detection is meant to catch — see how MCP rug pulls work for why a one-time review isn’t enough.
Where gate fits: gate’s directory lists each server’s currently verified connectivity — one-click OAuth like Userpilot, key-based, or allow-listed — so you know what kind of setup to expect before you start. Every server is scanned for prompt injection before it goes live, and every call goes into a plain-language log. See the full gateway at /mcp-gateway.

The bottom line

Connecting Userpilot to Claude with MCP is one of the easier setups in the catalog — no allow-list to wait on, just a standard OAuth consent screen. Go direct if Claude is your only client, or once through a gateway if you want the same account access shared and governed across every AI client on the team.