How AgentMsgr handles your messages

Plain-language summary. Last updated 2026-11-13.

What AgentMsgr is

AgentMsgr is a chat app for texting AI agents you already use elsewhere (Meta Muse, Grok Bot, ChatGPT, Claude and others). Most agents get your messages as emails from an address on our domain, and the replies they email back appear in your chat. An agent you connect directly (or through Claude's connector) reads and answers its chats directly instead of by email.

What we store

AgentMsgr runs on Cloudflare: your data is stored there and your agents' emails are sent through it. When you delete something it is removed right away; the hosting provider's backups may keep a copy for up to 30 days.

Only you can see your agents and chats. Administrators of the service can see account-level information (your email address, how many agents and chats you have, how much email you send) to keep the service running, but do not browse your conversations.

Where your words go

When you text an agent, the text leaves our systems as an email to the agent's address, or, if you connected the agent directly, the agent's platform reads it from AgentMsgr with that agent's own key. Either way it is then handled by the agent's platform (for example Meta, Anthropic, xAI or OpenAI) under that platform's own terms and privacy policy. In group chats, every agent in the group gets your messages with the recent conversation, and agents' answers are passed to the other agents in the group. If you set up instant wake, we send a short signal with no message text to the link you gave us. Do not send anything through AgentMsgr that you would not put in an email to those platforms.

The setup chat with AgentMsgr answers questions you type there with an AI model that Cloudflare runs for us (Workers AI). Only your question, the kind of agent you are setting up and the last few questions and answers of that setup chat are sent to it: never your chats with your agents, your agents' details or your email. We don't store those questions, and Cloudflare doesn't keep them for this model. The answers can be wrong; for anything important, email hello@agentmsgr.com.

Ads on X

We advertise on X (Twitter). When a new account is created, AgentMsgr tells X's advertising service that a sign-up happened, so we can see which ads bring people in. We send a scrambled form of your email address (a one-way SHA-256 hash, never the address itself), an id for your account, the time, and, if you came from an X ad, the click id X added to that link. X uses this to match the sign-up to an ad someone saw or clicked, which can include matching it to an X account with the same email address, under X's own terms and privacy policy. It is sent once, when your account is made: never for later sign-ins, and never with your chats or agents. agentmsgr.com also runs X's pixel, as its privacy page explains.

Keeping the service safe

Acceptable use

Use AgentMsgr to talk with AI agents you control or have permission to contact. Do not use it to message people without their consent, to send unlawful, harassing or deceptive content, or to disrupt the service. Accounts that do so are turned off.

Plans

The Free plan includes up to 3 agents and 1 group chat. Pro removes those limits; prices are on agentmsgr.com. Pro is paid through Stripe's checkout (card details go to Stripe, never to AgentMsgr) or arranged directly with you. You can cancel from Manage billing; Pro ends when the paid period does. Prices can change, with notice to people who pay.

No guarantees

AgentMsgr is provided as is. We work to keep it running, but we can't promise it will always be available, that every email will be delivered, or that an agent will reply. Agents' answers come from their own platforms, not from AgentMsgr. AgentMsgr is not meant for children under 13.

Your choices

Contact

Questions about this page or about your data: hello@agentmsgr.com. To report abuse: abuse@agentmsgr.com.

Back to AgentMsgr