Being a development kit, there's a small amount of construction required.
We have put together a detailed set of instructions to help get you setup. These instructions include the laser cut enclosure, 3D printed audio chamber and camera module.
The Mark II, including the Dev Kit, comes with 3 buttons and 1 switch. These are all located on the SJ201 board.
Action - the button located in the center of the LED ring. This activates the listener if needed.
Volume Up - the button on the front-right of the board when looking at the screen.
Volume Down - the button in the front-center of the board.
Mic hardware cut-off - a physical disconnect switch for the microphones. Switch left to disable the microphones, and right to re-enable them.
The SJ201 comes with 12 LEDs in a ring formation.
Currently there are two "dedicated" LED's for development purposes only (these will change as we approach a general release):
One led is green/red representing the current state of the microphone mute.
The second shows the general CPU temperature
Blue = cold
Magenta = warm
Orange = hot
Red = very hot
The LED ring has a number of states:
Blue and spinner - listener is activated such as when the wake word is detected.
Pulsing ring - device is thinking.
You can find detailed information on all components of the Mark II hardware on our Github.
Download the latest version of the Mark II operating system from:
Using a tool like the Raspberry Pi Imager, flash the downloaded image onto the USB drive provided with your Mark II Dev Kit. To get the most from this drive, be sure to plug it into the top left USB port which is blue indicating it is USB 3.0.
When you power on the device for the first time, Mycroft will walk you through setting up your WiFi and pairing the device to your Mycroft account. If you haven’t yet registered, you can create an account at https://home.mycroft.ai.
The software on the Mark II Dev Kit is still an early release. Please help us improve the Mark II by opening a bug report or feature request for anything that doesn't seem right.
There are a number of known issues that we are working through at the moment. These include:
The Wifi Connect application does not yet support hidden SSID's.
Workaround - ensure your network SSID is visible.
Cannot poweroff the device. The system currently fails on shutdown and eventually reboots.
Workaround - select reboot / poweroff, then wait 5 seconds before unplugging the device.
This also affects reboot, taking a long time to cycle - you can safely remove and re-insert the power cable after a 5 seconds to speed up the process.
Additional settings in on screen menu not yet implemented:
On screen Wifi setup
To enable secure access to your device, we use public-key cryptography rather than a default username and password.
In short this means that you generate what’s known as a key-pair. The pair is made up of two files - a public key and a private key. The public key is transferred to your device, and only your private key will be able to log in to it. Your private key is like an extremely secure password and you should never share this with anyone. If you want to learn more about the details, start here.
Open your terminal and run:
ssh-keygen -t rsa
You will be asked where to save your new key-pair
Leave this blank if you would like to use the default location
You can optionally add a passphrase. This is an additional layer of security that will require you to enter your passphrase when using your private key.
If you selected the default file path you will have two new files:
/home/user/.ssh/id_rsa is your private key - again do not share this.
/home/user/.ssh/id_rsa.pub is your public key - this can be shared.
If you read the contents of your public key it will look something like this:
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDSUOctoVJ5nwQO0I9c8gIK0ijYbOCZKdVmAu8jG0Znl2zvZLFYI7bGFbt/Rr8vjFVh4I2srFB52duirX53LtZF2ZUKMI+8ivaLk+pD7M0WL+sbG1jU2S/IdCUi7HmZ/fSp89UJb23i9Q9AINFdw+0spCjJSWB8+3wGQ4bPUSbNLCtsYv1anO+B26PIN5E9R0X84IVq+x41B1swvlUt15zfMwA7Jhl5fJgl6XzhNYcMdH/qp+r7Ij2d7NM9YR6/yva4/QLqzbPCoelxJLpVHKZ0ZLnhvYOoxRbLbU46AgTljGM3Z7rcvxm2Vl107ZojljuvF6cMXM4NU4peVimn5XOP [email protected]
This includes three things:
The encryption protocol used -
The public key - a seemingly random string
An identifier of the creator -
Now that you have your key-pair setup, you can add it to any of your Mycroft devices at: https://home.mycroft.ai/devices
Your Mark II device will then fetch it from your Mycroft account.
With your public key on your Mark II, and your private key remaining securely on your local machine, you can now SSH into the device.
You will now be logged in as the
mycroft user, with the virtual environment active. Your prompt should look like:
(.venv) [email protected]:~$
From here you can interact with the device as you would any other Linux system. Mycroft's standard tooling is also available including:
mycroft-cli-client - a terminal based graphical interface for the system
mycroft-msm - Mycroft Skills Manager - for installing, removing, and updating Skills
mycroft-msk - Mycroft Skills Kit - for quickly generating new Skill templates
mycroft-skill-testrunner - for running Skill integration tests
There are 3 areas that make up the operating system on your Mark II:
mycroft container - contains everything you would expect and is built on a base of Ubuntu 20.04.
awconnect container - contains the WiFi setup application and manages the network connections.
_pv_ (Pantavisor) initrd view - boots the system, manages the other containers as well as updates.
You can SSH into both of these containers and the initrd view by using the container name as the username. For example to SSH into
awconnect you would run:
Or to SSH into the Pantavisor initrd view you would run:
There is also a BSP layer that can be mounted for inspection on another machine. This contains the kernel, modules and firmware.
To create your first Skill run:
Then see our detailed Skill development documentation:
Mycroft Skills Manager (MSM) is a command line tool used to add, manage and remove Skills on any Mycroft installation. It can install any Skill listed on the Mycroft Skills Repository or from any Github repository.
The idle screen (also called a Home screen) on the Mark II is fully customizable. The default screen is provided by the Time Date Skill but you can create your own.
Here we have a simple example Skill that allows the user to set their idle screen to be an image from a remote url through the Skill's settings.
To switch between the available screens, pull down from the top of your screen to access the on device menu and select Additional Settings > Homescreen. As new options are added to Skills they will automatically show up in this list.
Before making changes to
mycroft-core on the Mark II Dev Kit it's recommended that you disable automatic updates in your device settings.
To return to a production state, it is recommended that you flash a fresh image of the Mark II OS.
Mycroft-core is installed at
Automatic updates can be disabled or re-enabled from the Device Settings in your Mycroft account.
By default, your Mark II checks for updates on a regular basis. When an update is available, it will apply it after a preset delay.
When requesting an update your device will fetch this directly from Pantacor’s servers. As we've always done, connections are initiated by the Mycroft Mark II. The server cannot initiate a connection with a device.
The Mycroft Mark II uses Pantacor to provide system updates.
Mycroft has partnered with Pantacor to provide a comprehensive and robust software life-cycle management solution. It is an open source solution that uses container technologies to securely and reliably maintain edge services on Linux devices.
For the Mark II this provides a very stable and resilient operating system and update service. If something goes wrong on your Mark II, the device will automatically roll back to a previously working state.
The team at Pantacor share our commitment to open source, privacy and security. Pantacor’s co-founders also have a strong reputation to back that up. Furthermore, as they are constituted in Europe, they must also meet the EU’s strict regulatory guidelines including the GDPR.
No, your Mycroft account handles everything and is the only account you need.
Pantavisor is open source software from Pantacor that manages the containers on your device and handles the update processes. You can find the source code for it here on Pantacor’s Gitlab.
In more technical terms, Pantavisor is a device-side initrd base system which assembles a userland made up of one-to-many containers. On the Mark II, it starts the awconnect container, providing networking and connectivity, and the mycroft container that contains everything that runs the Mycroft application including mycroft-core, mycroft-gui, Skills and all the dependencies for these.
For further information on Pantacor check out:
Pantacor provides a method of cloning the configuration or “factory state” of a device, or any pristine update state that is available for a device. This is the configuration and other static assets used to generate or update an image. It does not, and cannot, clone the contents of your device. It cannot read your filesystem. A “clone” does not contain any of your Skills, Skill settings, data or other files.
Device cloning would not be useful to perform on any end-users device. It is only useful during development to share modified system configurations with other developers.
Mycroft is opt-out by default. That means if you haven’t explicitly opted-in to share your data with us, then we do not keep it. Any queries sent to our back end services are processed and then deleted immediately.
General usage statistics, such as how many unique devices have connected to our servers in a given day are collected and kept in aggregate. This allows us to see general user trends, but again, if you haven’t opted-in we would not know whether any device from a specific account was even turned on any previous day.
In developer mode, the Pantavisor system provides an SSH bridge running on port 8222. This bridge allows you to enter a shell session into any running container, regardless of whether that container is running an SSH server by itself. Due to the developer-friendly nature of the Mark II, this is enabled by default however no access is allowed until an SSH key is added to the Mycroft management panel.
In previous devices like the Mark 1, we provided the ability to turn SSH on and off because they operated off a default username and password. Distributing an internet connected device with a default username and password providing shell access is a very big security risk. Hence the SSH service had to be disabled by default.
The use of a key-pair removes the need to disable the SSH server. Without a public key on the device, and the ability to connect to your device on port 8222, no one can SSH into it.
First, we recommend using the Pi that ships with your device.
If you need to switch this out, please note that there are some revision 1.4 Raspberry Pi's that ship with firmware that does not support USB boot. To update the firmware, you can flash the Mark II image onto a Micro SD card. This will automatically update your firmware to a supported version on first boot. From then on you can boot from USB or continue using the Micro SD card.
The Mark II is intended to be a device for consumers. As such we try to keep the system as lean as possible. How lean? Currently it is "no-installed-text-editor" lean!
mycroft container however is based on Ubuntu 20.04 so all of your standard tooling is available via the apt package manager.
The Raspberry Pi GPIO pins 1, 12 and 13 are brought out to J9 on the SJ201. This is next to the volume up button on the front-right of the board if you are looking at the device face on. GPIO pin 0 is also not used.