The Typeform MCP server lets you build an integration that works with Typeform accounts from an LLM client. It exposes form and automation building, contact management, and response analytics over the Model Context Protocol — an open standard for connecting AI assistants to external tools — so you can offer Typeform capabilities without building against individual REST endpoints.
💡 Just want to use Typeform from an AI client yourself? You don't need these docs — start at Connect Typeform to your AI with the Typeform MCP server or connect to the native MCP connectors in Claude or ChatGPT.
| Capability | Tools |
|---|---|
| Create and edit forms — fields, logic, publishing | forms-* |
| Build and publish automations with email, webhook, delay, and integration steps | automations-* |
| Manage contacts and lists, including bulk upsert and form field mapping | contacts-* |
| Query response analytics — raw submission data, aggregates, rankings, time series | insights-* |
See Supported tools for the full list, and Not yet supported for the capabilities that still need the REST API.
Connect an MCP client to:
https://api.typeform.com/mcpStreamable HTTP only — there's no SSE endpoint, and clients configured for SSE fail immediately. Authorization is OAuth 2.0, and your client is prompted on first connect.
To try it without writing any code:
npx @modelcontextprotocol/inspectorCall accounts-list_accounts first: nearly every other tool needs the account_id it returns.
EU-hosted accounts use different endpoints, and a hosted connector needs its redirect domain allowlisted first — see Core concepts and Build an MCP connector.
submit_feedback matters