You can connect Typeform to Claude using the Model Context Protocol (MCP), and once it’s connected, Claude can pull up a form’s latest responses or summarize what people actually wrote without you exporting a spreadsheet first. Typeform’s OAuth currently only admits clients the vendor has allow-listed in advance, so the setup has one extra wrinkle most “connect X” guides don’t. This guide covers what Typeform’s MCP server does, both ways to connect it — directly or through a gateway — and why form responses deserve a slower look than most data in this series.
What the Typeform MCP server actually does
Typeform publishes a remote MCP server at api.typeform.com/mcp. It exposes your account’s forms, responses, and insights as tools Claude can call directly, so prompts like these work against your real data instead of a CSV export you’d otherwise have to open:
- “How many responses did the event signup form get this week?”
- “Summarize the open-ended feedback on the last customer survey.”
- “Which question in the application form has the most drop-off?”
- “List everyone who answered ‘no’ to the eligibility question.”
Claude reads through the same account permissions your Typeform login already has — it’s querying live form and response data, not a static export.
The catch: Typeform’s OAuth is allow-listed
Several servers in gate’s directory — Slack, Figma, Vercel, HubSpot — currently only admit AI clients the vendor has allow-listed in advance, and Typeform is one of them. In practice that can mean the connection attempt fails at the authorization step with no obvious explanation, which reads like a broken setup when it’s really just a client Typeform hasn’t cleared yet. See the unauthorized client error, explained for why vendors do this and what the options are. If a client you use hits that wall, a static personal access token — if Typeform’s API offers one for the scopes you need — is usually the fallback worth checking before giving up on the connection entirely.
Option 1: Connect Typeform 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 Typeform’s MCP endpoint.
- Sign in with your Typeform account when prompted — if Claude isn’t on Typeform’s allow-list yet, the consent screen will say so rather than completing.
That’s enough if the connection goes through and Claude is the only AI client you use with Typeform. The tradeoff shows up once a second client or a second teammate needs the same access: each one repeats the same approval step, 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 Typeform (and everything else) to one gateway URL, then point every AI client at that single URL instead of repeating the setup per client.
With gate specifically, connecting Typeform looks like this: add gate’s gateway URL to Claude once, then pick Typeform from the Typeform server page in the directory and connect it the same way you would directly — OAuth if your client clears the allow-list, or a personal access token in the same form if it doesn’t. From then on, Typeform’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 forms
A Typeform account usually holds more than the one survey you had in mind when you connect it — every job application, every event signup, every customer feedback form across the account. A few things worth checking:
- Who actually wrote the responses? Unlike a product-analytics tool reading your own usage data, a form’s responses were typed by people outside your company — applicants, event registrants, survey respondents — who almost certainly didn’t picture an AI client reading their open-ended answers back later.
- Does a form collect anything sensitive? Application forms routinely ask for names, emails, and sometimes more. Know which forms in the account carry that before asking Claude to summarize “the applicant pool.”
- 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 or delete a live form rather than just read its responses.
- Who else on the team connects it? If more than one person wires up Typeform 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
- The OAuth screen won’t complete. That’s almost always the allow-list, not a misconfiguration on your end — see the unauthorized client error, explained.
- Claude can’t find a form you know exists. Forms don’t carry over between separate Typeform workspaces — check the connection is pointed at the right one.
- Tools disappear after a Typeform 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.
The bottom line
Connecting Typeform to Claude with MCP works the same way as any other remote server, with one extra step: your client needs to clear Typeform’s allow-list, or fall back to a personal access token if it doesn’t. 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 — and either way, remember that the responses on the other end were typed by people who never expected an AI to read them.