About niceform-mcp
niceform-mcp lets MCP-compatible clients create and manage niceform.ai forms, from drafting through response analysis and CSV export. It is for agents handling the full form lifecycle through a set of dedicated tools.
Key features
- 01Create forms from descriptions
- 02Preview and publish forms
- 03List and close forms
- 04Retrieve paginated responses
- 05Summarize responses
- 06Export responses as CSV
Use cases
- 01Build a form from a natural-language request
- 02Review a form before making it public
- 03Analyze submitted answers
- 04Export submissions for further processing
MCP support
Exposes nine niceform.ai form-management tools to MCP clients, including creation, publishing, response retrieval, summaries, and CSV export. It can run locally over stdio or connect to the hosted Streamable HTTP endpoint.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://www.niceform.ai/api/mcpclaude mcp add --transport http shirish-singh-niceform https://www.niceform.ai/api/mcp{"mcpServers":{"shirish-singh-niceform":{"type":"http","url":"https://www.niceform.ai/api/mcp"}}}Open in Cursor{"servers":{"shirish-singh-niceform":{"type":"http","url":"https://www.niceform.ai/api/mcp"}}}Open in VS Code{"mcpServers":{"shirish-singh-niceform":{"type":"http","url":"https://www.niceform.ai/api/mcp"}}}{"mcpServers":{"shirish-singh-niceform":{"serverUrl":"https://www.niceform.ai/api/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
https://www.niceform.ai/api/mcpAuth requiredTools
Details
From the project READMEniceform-mcp-server
MCP (Model Context Protocol) server for niceform.ai - Create AI-powered forms from any MCP-compatible client: Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, OpenAI Codex, Gemini CLI, and more.
Features
Nine tools, so an agent can run a form end to end:
- create_form - Generate a form from a natural language description
- publish_form - Make a form live and publicly accessible
- close_form - Stop accepting responses
- list_forms - List your forms
- get_form - Get a form's fields and status
- get_preview_url - Preview a form before publishing
- list_form_responses - Retrieve paginated submissions for forms you own
- summarize_responses - Summarize the responses to a form
- export_responses_csv - Export responses as CSV
Connect by URL (no install)
niceform.ai also hosts this server at https://www.niceform.ai/api/mcp (streamable HTTP). Send your API key as Authorization: Bearer <API key>.
claude mcp add --transport http niceform https://www.niceform.ai/api/mcp --header "Authorization: Bearer nf_live_your_api_key_here"
Cursor (~/.cursor/mcp.json):
{ "mcpServers": { "niceform": { "url": "https://www.niceform.ai/api/mcp", "headers": { "Authorization": "Bearer nf_live_your_api_key_here" } } } }
VS Code (.vscode/mcp.json):
{ "servers": { "niceform": { "type": "http", "url": "https://www.niceform.ai/api/mcp", "headers": { "Authorization": "Bearer nf_live_your_api_key_here" } } } }
Claude Desktop: use the npx setup below. More: https://www.niceform.ai/mcp
Requirements
- Node.js 18 or newer (the server runs locally through
npx, so there's nothing else to install) - A niceform.ai API key
- An MCP client that supports local (stdio) servers. See Supported clients.
Get an API Key
- Sign in to niceform.ai
- Go to Settings → API Keys
- Create a new API key
- Copy the key. It's only shown once.
Environment Variables
| Variable | Required | Description |
|---|---|---|
NICEFORM_API_KEY | Yes | Your niceform.ai API key (nf_live_...) |
NICEFORM_API_URL | No | API base URL (defaults to https://www.niceform.ai) |
Supported Clients
The server works with any MCP client that can run a local command, whichever LLM that client uses. In every example below, replace nf_live_your_api_key_here with your key.
| Client | Models | Config |
|---|---|---|
| Claude Desktop | Claude | claude_desktop_config.json |
| Claude Code | Claude | claude mcp add command |
| Cursor | GPT, Claude, Gemini and others | ~/.cursor/mcp.json |
| VS Code (GitHub Copilot) | GPT, Claude, Gemini | .vscode/mcp.json |
| Windsurf | GPT, Claude, Gemini and others | ~/.codeium/windsurf/mcp_config.json |
| Codex CLI / IDE (OpenAI) | OpenAI GPT | ~/.codex/config.toml |
| Gemini CLI | Google Gemini | ~/.gemini/settings.json |
| Cline, Zed, Continue, etc. | Any | Standard mcpServers JSON |
Browser-only apps (chatgpt.com, claude.ai, gemini.google.com) can't run local MCP servers. Use one of the desktop, editor, or terminal apps above.
Claude Desktop
Edit the config file, or open it from Settings → Developer → Edit Config:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"niceform": {
"command": "npx",
"args": ["-y", "niceform-mcp-server"],
"env": {
"NICEFORM_API_KEY": "nf_live_your_api_key_here"
}
}
}
}
Restart Claude Desktop afterwards.
Claude Code
claude mcp add niceform --scope user -e NICEFORM_API_KEY=nf_live_your_api_key_here -- npx -y niceform-mcp-server
Run /mcp inside Claude Code to confirm it's connected.
Cursor
Add the same mcpServers JSON as Claude Desktop to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project). It works with whichever model you pick in Cursor.
VS Code (GitHub Copilot)
Create .vscode/mcp.json in your workspace. VS Code uses a servers key, not mcpServers:
{
"servers": {
"niceform": {
"type": "stdio",
"command": "npx",
"args": ["-y", "niceform-mcp-server"],
"env": {
"NICEFORM_API_KEY": "nf_live_your_api_key_here"
}
}
}
}
Then use Copilot Chat in Agent mode.
Windsurf
Add the same mcpServers JSON as Claude Desktop to ~/.codeium/windsurf/mcp_config.json, then refresh Cascade.
Codex (OpenAI)
Add to ~/.codex/config.toml. The Codex CLI and IDE extension both use this file.
[mcp_servers.niceform]
command = "npx"
args = ["-y", "niceform-mcp-server"]
env = { NICEFORM_API_KEY = "nf_live_your_api_key_here" }
Gemini CLI
Merge the same mcpServers JSON as Claude Desktop into ~/.gemini/settings.json (all projects) or .gemini/settings.json (one project).
Other MCP clients
Any client that supports local (stdio) servers works. Most accept the mcpServers JSON above. Otherwise, use these values:
| Setting | Value |
|---|---|
| Command | npx |
| Arguments | -y niceform-mcp-server |
| Env | NICEFORM_API_KEY=nf_live_your_api_key_here |
Global install (optional)
If you'd rather not use npx:
npm install -g niceform-mcp-server
Then use niceform-mcp as the command, with no arguments, in any of the configs above.
From source
git clone https://github.com/Shirish-Singh/niceform-mcp && cd niceform-mcp && npm install && npm run build
Then point your client at node /absolute/path/to/niceform-mcp-server/dist/index.js.
Available Tools
create_form
Create a new form from a natural language description.
Parameters:
description(required): Natural language description of the formidempotency_key(required): A unique caller-generated key, preferably a UUID; reuse it when retrying the same creationstyle_preference(optional): One ofmodern,minimal,colorful,professional,default
Example:
Create a customer feedback form for a coffee shop with:
- Overall rating (1-5 stars)
- Service quality rating
- Coffee quality rating
- Free text comments
- Optional email for follow-up
get_preview_url
Get the preview URL for a form to see how it looks.
Parameters:
form_id(required): The form ID from create_form
publish_form
Publish a form to make it publicly accessible.
Parameters:
form_id(required): The form ID to publishname(optional): Override the form name
list_form_responses
Retrieve submissions for a form you own. Responses may contain personal data, so only request and expose fields needed for the user's task.
Parameters:
form_id(required): The form UUID or public IDstatus(optional):submitted,completed,incomplete, orall; defaults tosubmittedlimit(optional): Integer page size from 1 to 100; defaults to 25cursor(optional): Opaque cursor returned by the preceding pagesince(optional): ISO 8601 timestamp with offset; returns only responses submitted after it. Reuse the returnedlatestSubmittedAtto poll.
close_form
Stop accepting responses. Parameters: form_id (required).
list_forms
List the forms you own. Parameters (all optional): query (filter by name), limit (page size), cursor (from the previous page).
get_form
Get a form's fields and status. Parameters: form_id (required).
summarize_responses
Summarize the responses to a form: totals, per-field answer counts, option counts and min/max/avg. Parameters: form_id (required).
export_responses_csv
Export a form's responses as CSV. Parameters: form_id (required); status (optional): submitted, completed, incomplete, or all; defaults to submitted.
Example Workflow
You: Create a simple contact form with name, email, and message fields
Assistant: [Uses create_form tool]
Form created! Preview URL: https://www.niceform.ai/forms/abc123
You: Looks good, publish it
Assistant: [Uses publish_form tool]
Published! Share this URL: https://www.niceform.ai/forms/abc123
Development
# Watch mode (auto-rebuild on changes)
npm run mcp:dev
# Build once
npm run mcp:build
# Run the server directly
npm run mcp:start
Troubleshooting
"NICEFORM_API_KEY is not set"
Make sure NICEFORM_API_KEY is set in the env block of your client's MCP config. The key should start with nf_live_ or nf_test_.
"Invalid API key"
Your API key may have been revoked or expired. Generate a new one from the niceform.ai dashboard.
"Rate limit exceeded"
You've made too many requests. Wait for the retry period indicated in the error message.
Tools not appearing in your client
- Restart the client (or reload its MCP servers) after editing the config
- Check the config file syntax. JSON must be valid, with no trailing commas.
- Make sure Node.js 18+ is installed and
npxis on yourPATH. Some GUI apps don't inherit your shell'sPATH. If yours doesn't, use the full path tonpx(find it withwhich npx). - VS Code: use the
serverskey, notmcpServers
License
MIT
Links
- niceform.ai - AI-powered form builder
- MCP Documentation - Model Context Protocol
- MCP clients - Apps that support MCP
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.