Get Started

Up and running in two commands.

HiveMind installs on your own device. No servers to provision, no cloud account, nothing to operate. Run the installer, and your agents start sharing one memory.

1. Install

On each device you want in the Hive, run:

# install the hive-mind command
$ curl -fsSL https://raw.githubusercontent.com/projectmentor/hive-mind/main/scripts/installer/install.sh | bash

# set up this device (daemon, peers, device key)
$ hive-mind install
        

Runs on Linux, and on Windows 11 inside WSL. Multi-device sync runs over Tailscale. Mac and Android are on the way.

2. Use it

Any agent can read and write the Hive with one command:

# remember something
$ hv remember "The payments API rate-limits at 100 req/s" --tags api,payments

# find it later
$ hv search "payments"

# share with your other devices
$ hv sync now
        

Most of the time your agents handle hv for you, so you can step back and let them run.

3. Connect your agents

HiveMind already works with Claude Code, Hermes, and the MCP server for Claude Desktop.

4. Add another device

Install HiveMind on the new device the same way. It joins read-only at first (sterile): it reads the whole shared memory, but anything it writes is held back until you admit it. Its join-request surfaces automatically in the owner's next session, so there are no device IDs to copy around. From the owner device, run hv admit once to make the new device a full read-write member (fertile):

# on the new device: check this device's status
$ hv whoami

# on the owner device: promote the new device to read-write
$ hv admit
        

You only admit a device once. Identity survives reinstalls — uninstall with hive-mind uninstall --keep-identity and a reinstall resumes the same device, so you never have to re-admit it.

Removing HiveMind

One command takes this device back out. It stops the sync daemon, removes the hive-mind and hv commands, and strips the agent hooks:

# remove HiveMind and delete this device's data
$ hive-mind uninstall

# or keep your journal and keys so a reinstall resumes the same identity
$ hive-mind uninstall --keep-hive
        

With --keep-hive your journal and keys are saved to a backup folder. Tailscale and your shell PATH are left untouched.

Want the technical details?

Architecture, how it stays in tune, and the full CLI reference live on the developer site.

Go to the developer site