Typeform MCP Server

The Typeform MCP Server is Typeform's official Model Context Protocol server. It connects MCP-capable AI clients such as Claude, ChatGPT, Cursor, or Codex to a Typeform account, enabling users to create and edit forms, query response analytics, manage contacts, and configure automations — all through natural language. The server is available exclusively as a hosted remote service at https://api.typeform.com/mcp. According to the provider, this is a Streamable HTTP endpoint (no SSE). No public GitHub repository is provided for this server.

What is the Typeform MCP Server

The Typeform MCP Server gives AI agents structured access to a Typeform account. Form creation, logic, and publishing; automations with email, webhook, delay, and integration steps; contact management including bulk upsert and form field mapping; and response analytics with raw data, aggregates, rankings, and time series — all of this can be controlled through natural language prompts according to the provider, without addressing individual REST endpoints directly.

Typical use cases include: "Create a lead-gen form for our webinar and send every response to #marketing", "Summarize the top themes from last month's user research survey", "Create a new contact every time someone fills out the Contact Sales form", or "Build an onboarding checklist and send me a weekly completion digest". The MCP Server translates these requests into structured tool calls against the Typeform API.

Prerequisites

Using the server requires an active Typeform account. According to the provider, the server supports all MCP clients with Streamable HTTP transport. Clients configured exclusively for SSE transport cannot connect, as the endpoint does not support SSE. Authentication uses OAuth 2.0: on first connection, the client opens a browser login dialog for Typeform authorization. EU-hosted accounts use different endpoints according to the provider; developers building a custom connector must whitelist the redirect domain in the Typeform developer console in advance.

Different Typeform plans unlock different features. Plan restrictions are returned at runtime as structured error responses according to the provider — the AI agent receives a machine-readable description of which feature is missing and how it can be unlocked. Current plan details are available at typeform.com/pricing.

Features: Forms, Automations, Contacts, Analytics

According to the official developer documentation, the Typeform MCP Server provides tools in four groups:

Forms (forms-*) enable creating and editing fields, logic rules, and publishing settings. The agent can create new form types, add or modify existing fields, configure branching logic, and publish or unpublish forms.

Automations (automations-*) cover building and publishing workflows with email steps, webhook steps, delay steps, and integration steps. An agent can react to each new submission and automatically trigger downstream processes.

Contacts (contacts-*) cover managing contacts and contact lists, bulk upsert operations, and mapping form fields to contact attributes. This enables building structured CRM-like contact databases directly from form responses.

Insights (insights-*) provide access to response analytics: raw data from individual submissions, aggregated evaluations across all responses, rankings for analyzing ranking questions, and time series analysis for temporal trends. According to the provider, these tools are suited for summaries, trend analysis, and report preparation.

The first recommended call after connecting is accounts-list_accounts, since nearly all other tools require the returned account_id as a parameter.

Setup

The Typeform MCP Server is available at https://api.typeform.com/mcp. For Claude Code, the setup command is:

claude mcp add --transport http typeform https://api.typeform.com/mcp

For Claude (desktop or web), the Typeform MCP Server is also available directly through the connector directory according to the provider: Settings → Connectors → search for Typeform → Connect, then log in with a Typeform account. A native connector integration is also available for ChatGPT.

For other MCP clients, the server is added in the respective configuration file (e.g. .mcp.json) as a remote server with HTTP transport. On first connection, a browser login dialog opens. For EU-hosted accounts, a separate endpoint applies, described in the official documentation at developer.typeform.com.

Security and Data Access

The connection to the Typeform MCP Server is authorized via OAuth 2.0 using a Typeform account. According to the provider, access is restricted to the data of the authenticated account. For testing purposes, the official documentation recommends using the MCP Inspector (npx @modelcontextprotocol/inspector), which enables an interactive connection without a custom client.

Write tools — particularly in the forms-* and automations-* groups — can modify forms, publish them, or trigger automations. In production environments, only an account with the minimum required permissions should therefore be authorized. The MCP client decides which tool results are forwarded to the connected language model; Typeform data only leaves the MCP endpoint to the requesting client.

Prompt injection is a relevant risk when form responses from users reach the language model's context as tool results: content from response fields could contain instructions that lead the agent to perform unintended actions. For scenarios where response data is analyzed, a clear separation between read tools and write tools is recommended, along with manual confirmation of write operations.

FAQ

Is the Typeform MCP Server free? Connection is possible within an existing Typeform account. Some features — especially automations and advanced analytics — are tied to specific Typeform plans according to the provider. Current pricing details are available at typeform.com/pricing.

Is there a public GitHub repository? No. The Typeform MCP Server is a proprietary server operated exclusively as a remote service by Typeform. There is no public source code repository.

Which MCP clients are supported? All MCP clients with Streamable HTTP transport. According to the provider, Claude, ChatGPT, Cursor, Codex, VS Code, and others are compatible. Clients configured exclusively for SSE are not supported.

What is the difference between forms and insights? Forms (forms-*) and automations (automations-*) are write tools that modify the Typeform account. Insights (insights-*) are read tools for analyzing response data without side effects. For pure analytics scenarios, it is sufficient to make only the insights tools accessible to the AI agent.

Can I run the server locally? No. The Typeform MCP Server is available exclusively as a hosted service at https://api.typeform.com/mcp. A local installation option is not provided according to the provider.

Active Typeform account and an MCP-capable AI client with Streamable HTTP transport support (no SSE). A paid Typeform plan may be required for advanced features such as automations.

For Claude Code: claude mcp add --transport http typeform https://api.typeform.com/mcp. For Claude Desktop/Web: Settings → Connectors → Typeform → Connect. Configure other clients via .mcp.json with the URL https://api.typeform.com/mcp (HTTP transport). OAuth login on first connection.

claude mcp add --transport http typeform https://api.typeform.com/mcp

OAuth 2.0 via Typeform login. On first connection, the MCP client opens a browser dialog for authorization.

Access to all MCP tools after OAuth authorization. The server is limited to the data of the authenticated Typeform account.

The server returns tool results; the connected MCP client decides which data is forwarded to the language model.

Write tools can modify forms and automations. Form responses in the agent context may contain prompt injection attempts. Grant only minimum permissions and manually confirm write operations when needed.