Connect SafeLagoon to Claude, ChatGPT and Grok

SafeLagoon is a parental-control service for families. The SafeLagoon connector links your SafeLagoon parent account to an AI assistant you already use, so you can check your children's screen time and manage device rules in plain language.

Connecting is optional. You can keep using SafeLagoon in the app and on the web without it.

SafeLagoon MCP server URL:

https://mcp.safelagoon.com/mcp

Use this same URL in Claude, ChatGPT and Grok. Provided by SAFE LAGOON S.L.

What it does

After you sign in with your SafeLagoon account, you can ask the assistant things like:

Leo and Mia are example names. The assistant uses the children on your account.

You can also view top apps and daily totals, set blocking schedules, lock or unlock a child's device, and add or update a child profile.

Changes that affect a device (blocking, limits, schedules, locking, answering time requests) are separate actions from reading information. The assistant should ask you to confirm before it makes them. You can review or change the same settings in the SafeLagoon app at any time.

Requirements

• A SafeLagoon parent account, with at least one child device paired in SafeLagoon.
• A SafeLagoon plan that includes the assistant connector: SafeLagoon Premium, or Start your Premium trial. On a plan without it, the assistant will tell you the connector needs a plan that includes it.
• Some features depend on the child's device platform.
• A Claude, ChatGPT or Grok account. What you can add depends on your plan with that company (see each section below).
• Use the assistant in text chat. Voice modes may not load connector tools.

Connect SafeLagoon to Claude

Works on claude.ai and Claude Desktop. Claude connects to SafeLagoon from Anthropic's servers, so the connection is set up through your Claude account.

1. Open Claude and go to Customize → Connectors.
2. Click + Add, then Add custom connector.
3. Enter a name (for example, SafeLagoon) and paste https://mcp.safelagoon.com/mcp as the server URL. Click Continue.
4. Review the authentication settings Claude detected and leave them as they are. Under Authentication, choose Sign in now. If you see OAuth client, keep Use Claude's published identity (Recommended). Click Continue, then Add.
5. A SafeLagoon sign-in page opens. Sign in with your SafeLagoon parent email and password (and your one-time code if two-step verification is on), read the approval screen (see Signing in and permissions), and choose Allow access.
6. In a chat, click + at the lower left, open Connectors, and switch SafeLagoon on for that conversation.
7. Ask: "Which children are on my account and how much screen time does each have left today?"

Team and Enterprise plans: an Owner (or, on Enterprise, a person with the right role) adds the connector under Organization settings → Connectors → Add → Custom → Web. Members then open Customize → Connectors, find SafeLagoon and click Connect.

Claude's entry-level plan: Claude allows one custom connector per account.

To disconnect: go to Customize → Connectors, find SafeLagoon and choose Remove. To fully disconnect, also remove the AI Assistant device in SafeLagoon → Devices.

Claude's own guide: Get started with custom connectors using remote MCP.

Connect SafeLagoon to ChatGPT

Custom MCP servers are added on ChatGPT on the web (chatgpt.com), not in the phone app. You no longer need to turn on developer mode first. Availability depends on your ChatGPT plan and workspace settings. In a Business or Enterprise/Edu workspace, an admin may need to allow creating plugins with custom MCP servers. See OpenAI's guide for what your plan can do, including whether actions that change settings are available.

1. In ChatGPT on the web, open Plugins.
2. Select the plus button, then Add custom MCP server.
3. Enter a name (for example, SafeLagoon). Under Connection, enter https://mcp.safelagoon.com/mcp as the server URL.
4. Set authentication to OAuth.
5. Read the risk warning, tick I understand and want to continue, and select Create as a plugin.
6. Install the plugin. When prompted, select Connect and sign in to SafeLagoon (with your one-time code if two-step verification is on). Read the approval screen (see Signing in and permissions) and choose Allow access.
7. Start a new chat in text mode, type @, select SafeLagoon, and ask: "How much screen time does each of my kids have left today?"

If actions don't appear after you connect, open the plugin's details page and refresh it to pull the current list of actions.

ChatGPT asks you to confirm actions that change things. Check each request before you approve it.

To disconnect: go to Settings → Plugins, select SafeLagoon, open the ••• menu and choose Disconnect. To fully disconnect, also remove the AI Assistant device in SafeLagoon → Devices.

OpenAI's guides: Add custom MCP server and Connecting custom MCP servers in ChatGPT.

Connect SafeLagoon to Grok

xAI documents custom MCP connectors for Grok on grok.com. The steps below follow xAI's public docs. We have not end-to-end tested every Grok plan with SafeLagoon, so treat availability and exact screens as subject to your Grok account and xAI's current UI.

Grok support is new. If sign-in or connecting fails, contact SafeLagoon support.

1. Go to grok.com/connectors.
2. Click New Connector, then select Custom.
3. Enter https://mcp.safelagoon.com/mcp as the MCP server URL and complete the authentication.
4. Sign in to SafeLagoon in the window that opens (with your one-time code if two-step verification is on), read the approval screen (see Signing in and permissions), and choose Allow access.
5. Start a chat and ask: "Which children are on my account and how much screen time does each have left today?"

Grok Business and Enterprise: a team admin first adds the connector in the console (console.x.ai → your team → Grok Business → Connectors → + Add Connector → Other, then the server URL). Members then connect their own SafeLagoon account on grok.com/connectors.

Developers using the xAI API: the API has a Remote MCP tool that takes a server URL and a pre-issued token in an authorization header. It does not run SafeLagoon's interactive sign-in. To use SafeLagoon with Grok, add it as a custom connector on grok.com as described above.

To disconnect: on grok.com/connectors, find SafeLagoon in your connected list and click Disconnect. To fully disconnect, also remove the AI Assistant device in SafeLagoon → Devices.

xAI's guides: Connectors, Connector Management and Remote MCP Tools.

Signing in and permissions

When you add SafeLagoon, your assistant sends you to a SafeLagoon sign-in page. You sign in with your SafeLagoon parent account, and with a one-time code if your account has two-step verification. Then you see an approval screen.

The approval screen lists every permission the connector can request, with a short description of each, and you choose Allow access for all of them at once. You cannot pick a subset on that screen. The permissions it lists are:

• profiles:read
• profiles:write
• rules:write
• activity:read
• location:read
• alerts:read
• offline_access

Read the descriptions on the screen before you choose Allow access. These permissions are broader than the example prompts on this page. If you are not comfortable choosing Allow access for all of them, do not connect.

SafeLagoon uses OAuth. The AI company never receives your SafeLagoon password, and SafeLagoon never receives your Claude, ChatGPT or Grok password.

Example prompts

• How much screen time does each of my kids have left today?
• Show me this week's screen time report.
• Leo asked for more time. What is pending?
• Approve Leo's pending extra-time request.
• Block TikTok on Leo's phone.
• Set Mia's school-day daily limit to 2 hours.

Who it's for, and what it doesn't do

It is for parents and legal guardians who already use SafeLagoon to supervise a child's device. It works only with the children on your account. If you ask about someone else's child, it stays within your profiles and explains the limit.

It does not send or share information with other people. It does not start, sell or change a SafeLagoon subscription inside Claude, ChatGPT or Grok.

Privacy and security

• What the assistant sees. Only what is needed to answer your request, within the permissions you approved (for example, today's screen time for a named child). Once sent, that answer is held by the AI company under its privacy policy and settings. Check those settings with Claude, ChatGPT or Grok.
• What SafeLagoon receives. The requests the assistant makes to SafeLagoon (for example, "screen time for profile X today") and a technical identifier of the connected app. We do not receive your full conversation with the assistant, its memory or your other files.
• "AI Assistant" device. After you connect, a device labelled AI Assistant appears on your SafeLagoon parent account (SafeLagoon → Devices). That is the connector session.
• Check before you trust. AI assistants can make mistakes. For important decisions, check the SafeLagoon app. SafeLagoon is not an emergency service. If a child is in immediate danger, call your local emergency number.
• Only connect servers you trust. Claude and ChatGPT warn that custom connectors are not verified by them, so add only connectors from organizations you trust. Confirm the URL above before you add it.

SafeLagoon privacy policy: https://help.safelagoon.com/en/help-center/privacy-policy Terms of service: https://help.safelagoon.com/en/help-center/safe-lagoon-terms-of-service

Disconnect and revoke access

Do both of the following. Disconnecting in the AI app alone may not remove the connector from SafeLagoon.

1. In your assistant.
   • Claude: Customize → Connectors, find SafeLagoon, then Remove (or use the three dots next to it).
   • ChatGPT: Settings → Plugins, select SafeLagoon, open the ••• menu and choose Disconnect. To remove the plugin as well, choose Uninstall plugin.
   • Grok: grok.com/connectors, find SafeLagoon in your connected list and click Disconnect.
2. In SafeLagoon. Open your SafeLagoon parent account and remove the device labelled AI Assistant. Some AI apps (for example Claude) may not tell SafeLagoon when you disconnect, so that device can stay until you remove it yourself. Removing it ends the connector session on SafeLagoon's side.
3. Disconnecting does not delete data the AI company already received. Use that company's own tools for that.

Troubleshooting

The sign-in page doesn't open, or Claude, ChatGPT or Grok says it can't connect. Check that the URL is exactly https://mcp.safelagoon.com/mcp, with no extra spaces. If your browser blocks pop-ups, allow them for the assistant's site and try again. If it still fails, remove the connector, add it again and complete the sign-in. If it keeps failing (especially in Grok, where support is new), contact SafeLagoon support.

I connected, but ChatGPT shows no actions. Make sure you selected Connect and finished the SafeLagoon sign-in, then refresh the plugin on its details page.

The tools worked in text chat, then disappeared. Voice modes may not load connector tools. Go back to a text chat.

The assistant says I need a plan that includes the connector. The connector needs a SafeLagoon plan that includes it. Check your plan in your SafeLagoon account or contact support.

The assistant can't see my kids. Ask: "Which children are on my account?" If the list is empty, open the SafeLagoon app and check that at least one child profile and device are set up.

The assistant asked me to confirm a change. That is expected for actions that change settings. Read the confirmation carefully, or cancel and make the change in the SafeLagoon app.

I can't find "Add custom connector" or "Add custom MCP server". Availability depends on your plan with the AI company and, in a team workspace, on admin settings. See the vendor guides linked above.

Something looks wrong with permissions or data. Disconnect the connector in the AI app, remove the AI Assistant device in SafeLagoon, then contact support if needed. You can also change the same rules directly in the SafeLagoon app.

Support

• Support: https://help.safelagoon.com/en/help-center/how-to-contact-us
• Help center: https://help.safelagoon.com
• Website: https://safelagoon.com

SafeLagoon is provided by SAFE LAGOON S.L. Claude is a product of Anthropic, ChatGPT of OpenAI and Grok of xAI. These companies are independent of SafeLagoon.