Manage your forms from your AI assistant

Connect Claude or ChatGPT

No token to copy: you add the server address, the assistant opens an AwesomeFor.ms page, and you allow it. You can disconnect it at any time from Settings, section “AI assistants (MCP)”.

Claude (claude.ai, Claude Desktop)

1. Open Customize, then Connectors, choose Add, then “Add custom connector”.
2. Name it AwesomeFor.ms, paste the address https://awesomefor.ms/mcp and click Continue. Keep the advanced settings as they are, then click Add.
3. Click Connect, sign in to AwesomeFor.ms if asked, then allow access.

On a Team or Enterprise plan, an owner of the organisation adds the connector first.

ChatGPT

1. Open the connectors settings of ChatGPT and choose to add a custom MCP server.
2. Paste the address https://awesomefor.ms/mcp and, if asked how to authenticate, choose OAuth.
3. Sign in to AwesomeFor.ms when asked, then allow access.

Custom MCP servers depend on your ChatGPT plan and workspace settings: see OpenAI’s help article “Connecting custom MCP servers in ChatGPT”.

Advanced: personal access token

Claude Code, Cursor, VS Code and the other MCP clients can also connect with your approval. If yours does not, or for a script, use an access token.

1. Create an access token

Sign in, open Settings, then the “AI assistants (MCP)” section and open “Advanced: personal access token”. Name the token after the tool that will use it and copy it: it is shown only once. You can keep up to 10 tokens and revoke each one at any time.

2. Add the server to your client

The server address is https://awesomefor.ms/mcp and the token goes in the Authorization header. Replace afm_… with your token.

Claude Code

In a terminal:

claude mcp add --transport http awesomeforms https://awesomefor.ms/mcp --header "Authorization: Bearer afm_…"

Cursor

In .cursor/mcp.json (this project) or ~/.cursor/mcp.json (all projects):

{
    "mcpServers": {
        "awesomeforms": {
            "url": "https://awesomefor.ms/mcp",
            "headers": {
                "Authorization": "Bearer afm_…"
            }
        }
    }
}

VS Code

In .vscode/mcp.json; VS Code asks for the token once and stores it securely:

{
    "inputs": [
        {
            "type": "promptString",
            "id": "awesomeforms-token",
            "description": "AwesomeFor.ms access token",
            "password": true
        }
    ],
    "servers": {
        "awesomeforms": {
            "type": "http",
            "url": "https://awesomefor.ms/mcp",
            "headers": {
                "Authorization": "Bearer ${input:awesomeforms-token}"
            }
        }
    }
}

Other clients that speak MCP over HTTP (“Streamable HTTP”) work the same way: the address above, and the Authorization: Bearer header.

Ask

For example: “Create a membership form for my badminton club, with a page per household member, then publish it.” or “Summarise this week’s responses to my satisfaction survey.”

What the assistant can do

• list_forms: Lists your forms with their status, number of responses and public link.
• get_form: Reads the full structure of a form: pages, fields, options and conditions.
• create_form: Creates a draft form from a list of pages and fields.
• update_form_structure: Replaces the pages and fields of a form, and optionally its title and description.
• publish_form: Publishes a form so that it accepts responses, after the same checks as the editor.
• unpublish_form: Takes a form offline: it no longer accepts responses.
• list_responses: Reads the responses of a form, newest first, 50 per page, with the same columns as the CSV export.

The assistant cannot delete a form or a response, nor change your account or your subscription.

Limits and safeguards

• The same rules as in the editor: your plan’s field types and quotas, and the same checks before publishing.
• Forms are created as drafts. On an account less than a day old, or one we are still checking, publishing is done from the editor.
• Up to 60 requests per minute per token or connected application.
• Every change made through MCP appears in your activity log, and a revoked token or a disconnected application stops working at once.

Your respondents’ data

Responses read by your assistant are sent to its provider (Anthropic, OpenAI, etc.) under your own agreement with them. Before connecting a form that collects sensitive data, check that this provider suits you and that your respondents are informed.