toGrow

AI CONNECTION GUIDE

Connect AI to toGrow with MCP

Create and manage profiles through conversations in the AI app you use. Just add the server URL, log in to your toGrow account, then review the access permissions.

What is MCP and what do you need?

Model Context Protocol (MCP) is a standard that lets AI apps use tools from other services. The toGrow MCP connection provides tools to read profiles, check page addresses, create drafts, update content, and publish profiles.

  • Have a toGrow account ready. Don't have one? Create An Account first.
  • Use an AI app that supports remote MCP using the Streamable HTTP transport and OAuth with PKCE and dynamic client registration.
  • Make sure MCP connections are available in your AI app's version and plan. Apps that only support local stdio servers cannot use this URL directly.

You do not need to install Node.js, download code, or enter an internal API key for toGrow.

How to connect an AI app to MCP toGrow

  1. Open the connection settings in your AI app

    Look for MCP, Connectors, Integrations, or Add Server. Menu names vary by app. In ZCode, open Settings, then MCP.

  2. Add a remote server

    Enter “toGrow” as the connection name, select Streamable HTTP if asked, then copy this URL as the server address:

    https://togrow.id/mcp

    Select OAuth authentication or the automatic login flow provided by the app. Do not use this guide page's URL as the server address.

  3. Connect and log in through your browser

    Select Connect. When the login page opens, check that you are on the domain togrow.id, then log in with the account that manages your profile. Enter verification codes only on the login page, not in the AI conversation.

  4. Review and approve access

    Read the app name, callback address, and permissions on the consent page. The app name does not prove that the app has been verified. Select Allow Access only if you recognize the app and agree to its permissions.

  5. Return to your AI app and test the connection

    After authorization, return to your AI app. Start with a read request: “List the toGrow profiles I can manage.” If you have several profiles, specify which one to use before requesting changes.

Example prompts after connecting

Start with a draft and review it before asking AI to publish the page.

Read a profile

“Show my profile and summarize any information that still needs completing. Do not change anything.”

Create a new draft

“Check whether the address kedai-kopi-saya is available. If it is, create a draft profile named Kedai Kopi Saya. Do not add contacts or prices I have not provided, and do not publish it yet.”

Update, then review

“On the Kedai Kopi Saya profile, change the tagline to ‘Local coffee for every day’. Do not change anything else. Show me the result for review.”

Creating a profile through MCP produces a draft, not an automatically published page. After checking its contents, you can explicitly request publication. Changes to an already published profile may be visible to visitors; do not assume every change is a draft.

Understand permissions and protect your account

  • profile:read — read profiles you can manage and check page address availability.
  • profile:write — create drafts and change profile information.
  • profile:publish — publish profiles. Only request publication after reviewing the content.

Permissions are not limited to a single page: they cover profiles owned or managed by your account, including new profiles. This connection does not grant access to payments, passwords, or team member management.

Do not share OTPs, passwords, access tokens, authorization codes, or callback URLs in chat. Complete login and consent in the browser.

How to revoke an AI connection

Open AI Connections (MCP), log in if asked, then select Revoke Access for an app you no longer use. You can also find this page in Account Settings. Logging out of the dashboard does not revoke an AI app's OAuth connection.

Troubleshooting MCP connections

MCP or OAuth menu is unavailable

Check support for remote MCP, Streamable HTTP, PKCE, and dynamic client registration in your AI app's documentation. Menu names, plan availability, and transport support may vary. This guide does not guarantee compatibility with every AI app.

Opening the MCP URL in a browser does not show a page

The address https://togrow.id/mcp is an endpoint for AI apps, not a regular web page. A 401 response without a token means the endpoint requires authentication, not that the entire connection flow has succeeded. Enter the URL in your AI app's MCP connection menu.

Login, permissions, or tool loading failed

Make sure the server URL exactly matches the one above, complete login and consent in the browser, then reconnect from the AI app. A 404 may mean MCP is not enabled in that deployment. A 403 may mean permissions or connection policy denied the request. Do not try to fix this by entering internal secrets.

The profile I want to change was not found

Ask AI to list profiles first. Make sure the account used to grant access owns or administers that profile. Specify the correct profile before making further changes.