MCP Server for Forms: How AI Agents Can Create and Manage Your Forms
Every major form builder — Typeform, Google Forms, Jotform, Tally — requires you to open a browser, click through a UI, and manually build every form. In 2026, that's the slow way. stigmi is the first form builder with a native Model Context Protocol (MCP) server, which means you can use Claude, GPT, or any AI agent to create forms, manage questions, analyse responses, and run your entire form workflow — through natural language, without opening a browser.
What is MCP (Model Context Protocol)?
Model Context Protocol (MCP) is an open standard developed by Anthropic that defines how AI models communicate with external tools and services. Think of it as a standardised plugin system for AI agents. Instead of each AI application building its own proprietary integrations, MCP provides a common interface that any AI model can use to interact with any service that exposes an MCP server.
In practical terms, an MCP server is a service that exposes a set of "tools" — discrete actions that an AI can take. When you connect Claude Desktop to an MCP server, Claude gains the ability to call those tools. Claude decides which tool to call, constructs the right parameters based on your natural language prompt, and executes the action on your behalf.
The key insight is that MCP tools are typed and structured. They have names, descriptions, and parameter schemas. This means the AI can discover available actions, understand what each one does, and use them correctly — without any pre-programmed integration on the AI's side. A new MCP server gives any MCP-compatible AI model new capabilities immediately.
Technical note: MCP uses a JSON-RPC protocol over HTTP (Streamable HTTP) or stdio transport. The server exposes a tool list endpoint and a tool call endpoint. Clients (like Claude Desktop or Claude Code) discover available tools at startup and invoke them during conversations.
stigmi's MCP server: what it can do
stigmi's MCP server exposes 58+ tools across 16 categories, covering the entire form lifecycle. Here's a breakdown of what an AI agent can do via MCP:
How to connect Claude to stigmi via MCP
There are two ways to use stigmi's MCP server:
Option 1: Claude Desktop (stdio)
This is the simplest approach for personal use. Claude Desktop supports MCP servers via stdio transport. Here's how to configure it:
// In Claude Desktop settings (claude_desktop_config.json):
{
"mcpServers": {
"stigmi": {
"command": "node",
"args": ["/path/to/stigmi/mcp/dist/server.js"]
}
}
}Once configured, Claude Desktop will prompt you to authenticate with your stigmi account on first use (OAuth2 with PKCE — your credentials are never shared with the MCP server process directly). After authentication, Claude has access to all 58+ stigmi tools.
Option 2: HTTP MCP endpoint (production)
For production use or multi-user setups, stigmi exposes a Streamable HTTP MCP endpoint at https://stigmi.app/mcp. Any MCP-compatible client can connect with Bearer token authentication using a stigmi API token (available on Pro Plus plan).
# MCP tool call via HTTP
POST https://stigmi.app/mcp
Authorization: Bearer your-stigmi-api-token
Content-Type: application/json
{
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"name": "create_form",
"arguments": {
"name": "Customer Feedback Survey",
"description": "Post-purchase feedback for Q1 2026"
}
},
"id": 1
}This HTTP endpoint is suitable for integrating stigmi into automated pipelines, CI/CD workflows, or custom AI applications built with the Anthropic API or other AI providers that support MCP.
Real examples: what you can do with Claude + stigmi MCP
Here are concrete things you can ask Claude to do once stigmi MCP is connected:
"Create a customer satisfaction survey with a rating question, an NPS question, an open-ended feedback question, and a question asking if they'd recommend us. Make all questions required except the open-ended one."
Claude calls create_form, then calls add_question four times with the correct question types, labels, and required settings.
"Look at the responses to my 'Product Feedback' form and summarise the most common complaints."
Claude calls list_forms to find the form, then list_responses to retrieve submissions, then synthesises the open-ended text responses into a summary of recurring themes. No data export, no spreadsheet, no manual reading.
"My event registration form is getting a lot of drop-offs. Which question has the highest abandonment rate?"
Claude calls get_analytics and interprets the drop-off data by question, then provides specific advice on which question to simplify or remove.
"Duplicate my 'Q4 2025 Survey' form, rename it to 'Q1 2026 Survey', update the date question to say March 2026, and activate it."
Claude calls duplicate_form, then update_form to rename it, then update_question for the date question text, then activate_form. Four API calls, zero browser interactions.
Why no other form builder has this
As of early 2026, stigmi is the only major form builder with a native MCP server. Typeform, Google Forms, Tally, Jotform, and Paperform all have public APIs — but an API is not the same as an MCP server.
The difference is discoverability and AI-readiness. A REST API exposes endpoints — but an AI model doesn't automatically know which endpoints exist, what parameters they take, or when to use them. You have to build a custom integration: prompt engineering, function definitions, error handling, and context management. That's significant engineering work for every AI application you want to connect.
An MCP server is self-describing. When an MCP-compatible client connects to stigmi's MCP server, it automatically discovers all 58+ available tools with their names, descriptions, and typed parameter schemas. The AI model immediately knows what's possible and can use any tool correctly — no custom integration code required.
This matters because MCP is becoming the standard. Claude Desktop, Claude Code, and the broader Anthropic developer ecosystem all support MCP natively. As more AI products adopt MCP, stigmi's MCP server makes it compatible with an expanding ecosystem of AI tools by default.
Use cases for AI-driven form management
Who benefits most from connecting AI to their form builder?
Agencies and consultants
If you manage forms for multiple clients, MCP lets you spin up complete form setups in seconds from a Claude conversation — instead of manually clicking through the form builder for each client. Describe the form in plain English and Claude builds it.
Research and data teams
Create survey forms quickly, collect data, then use Claude to analyse responses without exporting to a spreadsheet. Ask Claude to find patterns, segment respondents, and summarise open-ended answers — all within the MCP conversation.
Developers building AI applications
If you're building an AI application that needs to dynamically create or manage forms (e.g., an AI assistant for event organisers, a customer success bot, an HR automation tool), you can use stigmi's HTTP MCP endpoint to give your AI agent form management capabilities without building a custom integration.
Operations and automation teams
Build pipelines where an AI model creates forms, monitors response rates, flags unusual patterns, and generates weekly reports — all triggered by a schedule or event, with no human intervention required.
Authentication: how it works
stigmi's MCP server supports two authentication methods:
- OAuth2 with PKCE — Used by the stdio transport (Claude Desktop, Claude Code). On first run, the MCP server opens a browser window for you to sign in to your stigmi account. The OAuth token is stored at
~/.stigmi/tokens.jsonand automatically refreshed. Your stigmi password is never exposed to the MCP process. - API token (Bearer) — Used by the HTTP MCP endpoint. Generate an API token in your stigmi dashboard (Pro Plus plan required), and pass it as a Bearer token in the Authorization header. Suitable for automated pipelines and production deployments.
Both methods are scoped to your account — the AI agent can only access your forms, not anyone else's. All MCP tool calls are logged and auditable in your stigmi account.
The full list: 58+ MCP tools across 16 categories
For developers and power users, here's a summary of the tool categories available in stigmi's MCP server:
| Category | Example tools |
|---|---|
| Forms | create_form, update_form, delete_form, duplicate_form, list_forms |
| Questions | add_question, update_question, delete_question, reorder_questions |
| Responses | list_responses, get_response, export_responses |
| Analytics | get_analytics, get_drop_off_data, get_completion_rate |
| Endings | create_ending, update_ending, delete_ending, list_endings |
| Logic rules | create_logic_rule, delete_logic_rule, list_logic_rules |
| A/B testing | create_variant, activate_variant, get_variant_stats |
| Public forms | get_public_form, submit_response |
| Annotations | add_annotation, list_annotations |
| Tickets | create_ticket, list_tickets, reply_to_ticket |
| API tokens | create_token, list_tokens, delete_token |
| Payments | get_payment_status, list_orders |
| Templates | list_templates, create_from_template |
| Admin | list_users, update_user_plan (admin only) |
| Health | get_health_status |
| OAuth | authorize, token_exchange, revoke_token |
Full documentation is available at stigmi.app/api-docs/mcp, including parameter schemas for every tool.
Where MCP is heading in 2026
MCP was released as an open standard in late 2024, and adoption has accelerated rapidly. By early 2026, hundreds of services have published MCP servers — from databases and code editors to CRMs and project management tools. The pattern is clear: MCP is becoming the standard way to give AI agents access to real-world tools.
For form builders, MCP unlocks several valuable workflows that weren't possible before. An AI agent can create a form based on a brief, activate it, monitor responses in real time, flag anomalies, generate weekly summaries, and close the form when a quota is reached — all autonomously, without any human interaction in the loop.
As AI models become more capable at planning and multi-step execution, the value of MCP-connected services increases. stigmi's early investment in MCP means that as new AI tools adopt the protocol, they immediately work with stigmi — without any additional integration work.
Related
Let AI manage your forms
Create a free stigmi account and connect it to Claude Desktop in minutes. 58+ tools for creating, managing, and analysing forms — all via natural language.