We need a Dinkum Dev-Community ASAP

My Mark II is coming in January, and I’m excited to be able to start playing around with it. But I can also read a room: obviously lots of people who paid money for this hardware device are disappointed by the state of the software on launch, and expected something usable right-out-of-the-box. From a usual commercial standpoint, where a product is evaluated for value-for-money, the value is all locked up in unrealized potential right now—as it stands the Mark II is a talking alarm clock. I’m a hobbyist/hacker, I never expected it to work out of the box, but I sympathize with people who wanted a working Siri/Alexa competitor this Christmas.

The absolute most important thing that needs to happen right now is community organization of development guides, tools, tutorials, videos, and news and updates for Dinkum-specific skills.

When Steve Jobs learned that the first Inside Macintosh development guides for 3rd-party devs were going to be delayed in printing, he found a phonebook publisher to pump out the cheapest, flimsiest pieces of crap he could so that the software development could start.

There are two sets of developer docs that I can find right now. The best ones with all the good tutorials are now out of date, since they’re for the old Mycroft core. The new ones are basically empty.

To get started with my Mark II, I need to know what’s changed in all the tutorials from the old docs. And then everyone who is developing, both in house and from the community, needs to start sharing tips and setups and instructions and tutorials that are Dinkum-specific. Look at any other Free Software project: they have constant changelogs, community engagement, development drives, etc. Code sprints, video series, etc. That’s whats going to get the Mark II into the device we all want to use and own.

In my honest opinion, while I understand many people are frustrated with the lack of skills for Dinkum, I think it’s counter-productive to do anything except encourage more development, and counterproductive to ask for anything but the resources to aid and broaden development from the company. The fact that they are shipping hardware is incredible, and a very, very difficult accomplishment. The software is set back and regressed, which is frustrating, but it can be brought up to speed.

The sooner we all focus on encouraging development and helping each other by building the tools, resources, and community spirit to support development, the sooner we can have the device we all wanted.


Going to be the elephant in the room.

All of that already happened and was there years ago… Just got ignored or at least neglected when the people that really understood FOSS left.

The good part is though, it is still “here” just no longer under the Mycroft umbrella.


Are you switching to Neon or OVOS?

I am the buildroot developer of OpenVoiceOS. :wink:

1 Like

Well that answers that question! I am gonna play with both Neon and OVOS over the next couple of weeks. I hope they work better than MycroftOS.


I’m just checking your GitHub out now—don’t want to get too off topic but I take it you don’t like the AGPL, since your timeline points out the date the Selene core was released with “a viral license.” At any rate, it’s clear that OpenVoiceOS stuff can’t be up-streamed back to Dinkum. That’s all well and good; it’s the nature of open products that forks occur and that users can choose the software they’d like to run, so all the best with your project.

I dropped money on a Mark II which isn’t arriving until January, and only now am I realizing the state of affairs regarding the software. I feel my investment is preserved with the continuation of the company which produced and maintains the hardware, and that company needs to get their software up to snuff fast.

I know how hard it is for a Free Software project to actually ship a hardware product—arranging parts and manufacturing and shipping and FCC licensing and all the rest is an insane enterprise, and since all those business relations and skills and certifications are vested in the Mycroft.ai company, that company needs to stick around. That investment in infrastructure is the bulwark against the other proprietary giants in this field—otherwise this becomes a bunch of nerds playing with Raspberry Pis.

They can’t just sell hardware and then let the community work on 3rd party replacements for their stock software—that’s not viable. So they need a Dinkum Free Software development community. If you’re saying they lost their community, then their focus needs to be on seeding and fostering a new Dinkum-centered one, right now.


Hey there, ovos developer here, I just want to clarify a couple things about ovos:

re licensing:

that is poor wording in the readme, we are not against GPL, but it is not compatible with our licensing policy, so OUR projects all have that distinguishing feature, which that “viral license” comment wanted to highlight but did poorly, more about out license choices here License - Documentation

re not being upstreamable:

we go to great lengths to ensure OVOS stuff is compatible with mycroft-core, the problem is that dinkum itself is incompatible with 95% of 3rd party software out there made for classic core.

except for ovos-core and ovos-lingua-franca we consider our projects a subset of the mycroft community, not a fork and not tied to ovos-core for the most part. most things are standalone modules that just connect to the messagebus, when possible we make ovos-only features optional to ensure things work in classic core (even if in a limited fashion)

as for actually upstreaming instead of making things compatible, mycroft posts in this forum have mentioned more than once that community giving unwanted feedback was one of the reasons dinkum was done behind closed doors, PRs have also been neglected for years before that, sometimes with comments such as “we have no time to review PRs we have not asked for”. Hard for people to continue trying to actively help with things other than minor bugfixes under these circumstances…

There are years of work done from the community for classic core, all that stops working in dinkum without warning, dinkum also works only in the mk2 which means a 500$ barrier to adoption, its no longer AI for everyone that runs everywhere.

All that said, I very much agree with your comments and general stance, I hope mycroft starts engaging the community again and gives us a clear roadmap so we can all collaborate, but for that we need to know what Mycroft wants to happen in the first place! Communication has been very unclear, i start to think that 3rd party skills for Dinkum are not only unsupported but unwanted as part of a misguided quality control attempt?


It all started with MycroftOS. This explains it.


Please play with both NEON and OVOS and report back your findings. Both negative and positive.

You are really helping a lot of friends / projects.


1 Like

I documented my findings about differences between dinkum and classic core, this should be a good starting point for anyone trying to understand what is different from dinkum and classic



A few things i think mycroft needs to do if this is to be a FOSS project with an active community

  • have a contributing guide
  • use a public issue tracker (github…), commits keep referencing a private Jira ticket number, if mycroft wants community help then we need to know what issues to fix…
  • have a roadmap, just like knowing what issues to fix we need to know where you want the project to go!
  • start using PRs and proper code review, the community needs to be part of the workflow, doing code archeology and commit inspection under dev branch doesnt work well, we are missing out on all the github goodies such as comments and side by side diffs documenting the changes that are happening

essentially it all boils down to communication, the most frustrating part is that some years ago mycroft had all this nailed down and an active community, either a mycroft person or a community member would answer any questions really fast, now neither happens which signals that the knowledge is gone from the collective hivemind


I started a repository for converted skills, if anyone wants to contribute. I got the mark II and wasn’t exactly expecting to spend a ton of dev time on this but I’ll review PRs and test things on my machine if it means we have more functionality on dinkum.


I don’t plan to convert my skills or build new Docker images.

I have been really disappointed by the team during the last year and I really think that Dinkum was a bad move.


It may not have been the wrong move for the long term. I was in a reddit discussion with the dinkum dev(no longer with Mycroft, btw) and he said that they moved to it because they couldn’t get a production quality unit out of the classic-core code.

However, I feel that they should have informed us before shipping and given us 3 options:

  1. If you are ok with tinkering, we can ship it to you now.
  2. If you are ok with waiting, we expect to have skills ported and the skill installer working by X date.
  3. If you don’t like option 1 or 2, click to process a refund.

I am just getting into Mycroft with the arrival of the Mark II. I have only been idly following things over the last year. I get the frustration if a bunch of prior work is basically not usable and I’d probably be in the same place. I don’t have the prior investment that a lot of people do, so I’m starting from the ground floor and this doesn’t fall to short of my expectations for an early adopter product. Though I was hoping skills would be more abundant and easier to install.


I conversed with the NEON dev last night and he pretty much echoed your sentiments.

It really is too bad. This is one of the projects I had high hopes of becoming a success. I had planned on buying one for each of my parents and my sister so they can get away from Amazon devices but I cannot, at this stage, purchase one for them. It is not consumer ready.

I will gladly tinker with it whenever I find some time but they do not have the skills to do so and I don’t want to have to maintain 4 units - 3 of which would be several hours away from me.

I fully agree @clintonthegeek . We must have a Dinkum community development effort that is directly connected to Mycroft AI’s development in order to have a successful platform to build off of. I’m quite interested in the other efforts and want to experiment with them as well, but a Dinkum / Mycroft community effort is necessary. I will check out @jmillerv 's repository.

1 Like

It’s not much at the moment. I just put a working port of hello and bark on my first day. I’m working on getting the Sing skill working because I think it’ll be 1) funny and 2) a degree more complicated than the first ones I ported but not by much.

1 Like

Thinking on it, what I want to do is open an issue for each skill in the marketplace that isn’t working. Then determining if they’re possible or worth it to port over. If they’re not possible or worth it, close the issue.


For the people that want a (Dinkum) community development group, just head over to the Mattermost server and join the “United Skill Writers” group.


(Like I said, it is/was already there. Just neglected. Feel free to give it another try)


Thanks for the heads up. I wasn’t even aware it existed.