Starting

Now that the installation is complete, all that is left to do is to run Mycroft. Start Mycroft with:

cd mycroft-core
./start-mycroft.sh debug

These simple commands will initialize all of Mycroft's core services (the microphone listener, the core skills, etc.) and will bring up an interface that you can use to troubleshoot the assistant and perform a limited set of basic commands (Figure 3). Setup, however, is not complete – and this fact reveals one of Mycroft's major caveats.

Figure 3: Mycroft's CLI interface for troubleshooting.

The Cloud

For Mycroft to actually function as a virtual assistant, it must be linked with Mycroft AI's online account service [9]. Opening an account is relatively painless, requiring a Google or Facebook account or a conventional email and password combination. Through the website, you can then pair the desktop client with the account service using a pairing code that you can acquire by asking "Hey Mycroft, pair my device." After this process is complete, Mycroft gains access to its core skills, such as reporting the weather and answering easy questions. The website also allows you to adjust some of Mycroft's settings, such as its voice, units of measurement, and more (Figure 4).

Figure 4: Mycroft Home's website allows you to change some of Mycroft's settings to tailor the system to your tastes.

Mad Skillz

Mycroft's modularity is one of its strong points. Mycroft AI encourages developers to write new skills for its assistant [10], and users are free to add or remove skills as they wish. Mycroft AI has created a convenient script that makes it easy to add a skill [11]. To download this tool is as simple as typing the following command in your terminal:

git clone https://github.com/MycroftAI/msm.git

To install a skill using the Mycroft Skills Manager (msm) script, all you have to do is:

cd msm
sudo -H ./msm install <skill name>

Just a tip – msm will require that you have the python3-dev, python3-setuptools, and python3-pip packages installed.