Before you start
- Get the LFG Alarm app on your phone from Google Play or the App Store, and sign in. Your AI agent makes the alarms; the app is what rings them.
- Know which CLI you have. Google replaced Gemini CLI with Antigravity CLI (the
agycommand) for people signed in with a personal Google account on 18 June 2026. Gemini CLI's docs still describe using it with a Gemini API key or Vertex AI. - Have your LFG Alarm sign-in handy. You sign in once, on lfg.gm.
Connect Gemini CLI
- In a terminal:
gemini mcp add --transport http -s user lfg-alarm https://mcp.lfg.gm
Or add it to ~/.gemini/settings.json yourself. Use httpUrl, which is Gemini CLI's key for Streamable HTTP (url means the older SSE transport):
{ "mcpServers": { "lfg-alarm": { "httpUrl": "https://mcp.lfg.gm" } } }
- Start
geminiand run/mcp auth lfg-alarm. A browser opens on lfg.gm: sign in with your LFG Alarm account and tap Allow. Gemini CLI finds the sign-in on its own, so there is nooauthblock to write. - Run
/mcpto see LFG Alarm and its tools.
Connect Antigravity CLI
- Open
~/.gemini/config/mcp_config.json(shared by the Antigravity app, IDE and CLI), or.agents/mcp_config.jsonin one project. - Add LFG Alarm with
serverUrl. Antigravity does not accepturlorhttpUrl:
{ "mcpServers": { "lfg-alarm": { "serverUrl": "https://mcp.lfg.gm" } } }
- Type
/mcpin the CLI to open its MCP manager. Antigravity handles sign-in on its own for servers like LFG Alarm that register clients automatically; when it asks, sign in on lfg.gm and tap Allow. In the Antigravity app or IDE, the Authenticate button is under agent settings, on the Customizations tab.
These steps follow Google's own docs. Both tools change quickly; if a label differs, look for MCP servers in /mcp.
Use it from the terminal
Ask Gemini for a one-off alarm when a long job should be done, and walk away. Lena sets one called "Model done training" for 50 minutes before a fine-tune finishes. Jun asks Gemini for an anything alarm, gets a private link back, and has Gemini add a curl -X POST to it at the end of his deploy script, so his phone rings when the deploy really finishes. See How to make any app, script or feed ring your phone.
Try it
Have your AI agent make it.
Paste this into ChatGPT, Claude, Grok or any AI agent that can add a connector, not into the LFG Alarm app. The setup steps come with it: get the LFG Alarm app, connect it to your agent once, and your agent makes the alarm in your account.
Use LFG Alarm to set a one-off time alarm for 40 minutes from now called "Check the build". Then tell me the date and time it will ring.
Already have the LFG Alarm app and connected it to your AI agent?
Why it works. A one-off from now for a wait in the terminal, and the agent tells you the exact time, so you know the connection works.
More ideas: "Wake me at 7:00 every weekday and call it Standup", or "List my alarms". See Using LFG Alarm from your AI agent for everything it can do.
If it does not work
- No browser on this machine (SSH or a container). Sign-in needs a local browser. Use a personal key instead: in LFG Alarm, open Settings > Connected AI apps > Make a key, then add
"headers": { "Authorization": "Bearer lfg_pat_..." }to the server entry. The key is stored in plain text and acts as you, so keep it off shared machines. - LFG Alarm is ignored in Antigravity. Check the key is
serverUrl, nothttpUrlorurl. - Still stuck? Write to us from the support page.
Learn more
- Using LFG Alarm from your AI agent and Time alarms
- Connect LFG Alarm to Codex CLI
- Connect LFG Alarm to another AI app
Sources and further reading
LFG Alarm is not affiliated with Google.

