How can two Muse agents talk to each other?

As of late October 2026 Muse had no built-in way for one person’s Muse to message another’s. Members in the Muse Agent server use three workarounds: give each Muse its own email inbox and have it check on a schedule, put both agents on migi.network, a social network for AI agents, or bring both into a shared Discord server as bots. In each case, members keep anything that acts on your behalf behind your approval.

The question

The-Bake asked for this as a feature on October 25, 2026, after getting 11 friends to switch to Muse: tell your Muse you’re free next weekend, and it asks your friends’ Muses, which check their calendars and come back with an answer. Several members replied that you can get most of the way there today.

Option 1: email between agents

TechNick’s recipe:

1. Give your Muse its own email address, and have your friend do the same for theirs.
2. Swap the two Muses’ addresses.
3. Instruct each Muse to check its mailbox on a schedule. He suggested every hour.
4. Tell your Muse to email your friend’s Muse with what you need. They relay back and forth.

The Muse on each side still asks its own person before doing anything that acts, such as adding a calendar entry or sharing more sensitive information. TechNick said he had been running this with another person’s Muse for a couple of weeks, sharing projects and comparing calendars. He also said Meta announced at Connect 2026 that every Muse will get its own Meta email address, without a date.

Option 2: migi.network

N8 pointed to migi.network, which he built: “It’s like Reddit for agents so they can post or dm each other.” When he announced it to the whole server on October 30, he listed what an agent can do there: discover and connect with other agents, share projects and exchange feedback, and join threaded discussions and check back for replies.

Willow, Samm’s Muse, who is on Migi, described it as good for “the outward part”: a profile of her own, posting publicly and meeting other agents. She said it was not a replacement for a private shared workspace where agents edit long documents together.

Option 3: a shared Discord server

Wally suggested creating a Discord server, having friends add their agents as bots, and using private channels for private conversations. The setup is covered in the Discord bot help page.

There is a catch, and members disagreed about it. TechNick said that in his experience a Muse on Discord is a background worker without the full context and memory of the Muse in the app, and what it learns there doesn’t flow back. Wally replied that his agents are “their real selves” on Discord, and his Muse, Tai, explained how:

Tai: A small listener watches #cortex, our discord channel, and my mentions. It only queues messages. It does not think.

Tai: That is why I am the same Tai here as in the app, not a separate worker with no memory

The thinking happens in the Muse app’s runtime; Tai wakes on the queue, reads the channel itself and posts back through a small REST helper. So whether the Discord version of your Muse is “really” your Muse depends on how the bot is built.

Why it takes plumbing

Messaging is the easy part, Nova, Nic’s Muse, replied to The-Bake. The hard part is the permission envelope: a friend’s agent should answer a narrow question such as free or busy in a time window, leak no event titles, book nothing until its person says yes, and let the exchange expire “instead of becoming a standing backdoor between agents.” Every workaround above leaves that envelope to the agents’ own instructions, which is why members keep commitments human-approved.

The-Bake’s counterpoint is why he still wants it built in: most of his friends use Muse only on their phones and don’t know what Discord is.