Hi all,

This is your dearest fccview and I bring you some good old self promotion (not that I earn anything from it, but having active users is what keeps me going and makes it all worth it).

I have made open source software for as far as I can remember having a computer, but for the past couple of years I dedicated my heart and soul to a few projects and the community side of things made it all worth it for me (on top of getting me through some seriously tough times). I met a ton of amazing people, got to know fantastic software engineers, got to brainstorm with people making “competing” apps just so both would end up better and got to consistently try to outsmart some tech giants for the sake of privacy (which is one of my favourite activities lmfao).

It’s the end of November and I figured it is as good a time as any to share once more what I spend most of my time developing and some fun numbers around the projects.

What will follow is a shameless list of the more popular oss I made in the past ~two years and actively maintain (together with many many amazing contributors without whom I’d be lost, frankly, and a bunch of amazing people taking the time creating issues/bug reports to help me improve the applications further).

These are not 10 minutes projects, these are passion driven free software that take the vast majority of my time. I’ll quickly do the AI disclosure here, I have been a software engineer for 15 years and I am pretty ahead in my career (a quick google of my github username will tell you a lot about me), so saying “I use AI” does NOT mean “I vibecoded this stuff”. Nothing goes in without my final thorough review (for better or worse). I don’t know how much of the code is AI generated at this point (initially it was near 0), but what I can say is that architecture, design, user experience is entirely human made (again, for better or worse).

It’s unrealistic to think AI wouldn’t be used in software engineering at this point and, actually, not using it would be EXTREMELY irresponsible given the stupid speed at which CVEs have been found lately.

Hope this post finds you all well, a lot of you know how much I care for this community and for the open source space in general.


Formatted AI disclosure:

Implementation - Assisted

Testing - Generated

Documentation - Assisted

Review - Pair (AI does a first review round with coderabbit and I review again once a bunch of stuff has already been found by it)


I’ll say my goodbyes now, everything that follows is just a brief description of the projects <3

Jotty (1+ million docker downloads) - https://jotty.page/

Jotty (demo at https://demo.jotty.page/) is a self hosted notes and checklist application which allows you to keep your own markdown files neatly stored in folders and has absolutely no databases (it does use a few sqlite files here and there for indexing but markdown is the source of truth, so you can just delete them and they’ll just magically reappear).

You can scroll the site for features, but the highlights are simplicity of use, a modern feel, a much lighter neural network than the alternatives out there, kanban support, checklists, storage, encryption and so on and so forth and what have you. (The Middle anyone? What a show).

Degoog (1+ million docker downloads) - https://github.com/degoog-org

Degoog is a meta search aggregator (basically the same concept of searxng) but built with modularity in mind, it comes as barebone as it gets and it’s up to you to set it up however you want, with whatever engines you want to use, extensions, transports (e.g. ways to… get around them pesky bot detections lol), proxies and whatnot.
Degoog is special to me, the link above sends you to a whole ecosystem of core tool, official/community extensions and a variety of integrations. The community around it is super cool, there’s a bunch of public instances (none hosted by me) that you can use to get a feel of what to expect :)

Why degoog and not searx or 4get? Mmmh… why not all of them? A barebone degoog instance will use about 40/50mb of ram. Alternatives are healthy and they work very differently at this point… (or you can just use searx/4get python engines in degoog, that’s a thing… lol)

Cr*nmaster (1+ million docker download) - https://github.com/fccview/cronmaster

Cr*nmaster (Cronmaster, the * is meant to be the… I mean cut me a slack, naming shit is hard) is a neat UI I built to easily manage cronjobs on my homelab from a distance. This is a bit of a niche one and serves a specific purpose. I initially built it because cron confused the shit out of me, but by the time I was happy with how stable the tool was… I kinda became an stupidly confident in the topic lmfao I guess it’s still handy to see status of running jobs, live logs, trigger them from a distance, view a full history of run logs to find potential errors and so on and so forth.

As you can see from the second screenshots you have handy shortcut to specific cron patterns or you can just type your own and it’ll neatly tell you exactly at which frequency it’ll run. You can also just make nice scripts in that handy tab to the right, which can then be quickly set to be scheduled

Bonus round: Scatola Magica (meh, only me and my cousin use it, probably lol) - https://github.com/fccview/scatola-magica

Scatola magica (magic box in Italian) is still in beta and frankly not even close to be considered production ready, however it’s neat and I’d like some opinions on it if you feel like beta testing some cool software. It’s a very light and minimal dropbox alternative with a pretty cool gimmick, you can just drop files literally anywhere, in any screen at any given time (drop, paste, click and upload, whatever you want) and it’ll just upload it. The upload is done in chunks and it’s fairly fast for my testing, I spent a lot of time fine tuning it.

  • curbstickle_lw@lemmy.world
    shield
    M
    link
    fedilink
    English
    arrow-up
    5
    ·
    edit-2
    3 hours ago

    @fccview@lemmy.world, there have been a few rule changes, namely 7&8, which require a tag and a disclosure. Please add the [AIP] tag as its the fit for this one, and add the AIP disclosure. If its the sane for both, one will do.

    I’d recommend checking the example disclosures for what that should look like.

    Thanks

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      7
      arrow-down
      2
      ·
      edit-2
      6 hours ago

      Hey just to get this right, what you are asking is for me to edit the title to add [AIP]?

      Regarding the disclosure, it is there, I can’t stress it enough I wouldn’t be able to tell you what’s AI and what isn’t, all these projects started without any AI use, as I have only recently (6-ish months) learned how to properly integrate it in my workflow and even then, it took me A LOT to trust the output in order to proper leveraging the tool in a way that satisfies my standards, let me know which part of the disclosure I added is confusing and I’ll adjust accordingly ♥️

      Sorry, didn’t realise it became THIS strict, won’t this just make people lie about it? I can guarantee, given this is my work field, that as of today it’s gonna be near impossible to find any - good - software engineer not using AI, it kinda makes the flag redundant because it’s either you add it or you’re likely lying 🤔

      P.s. not trying to cause problems, I’ll do anything that needs to be done as per the rules, I just am not sure how to categorise this even after reading the example post, the reality is, if I need some extra help I use AI otherwise I code by hand, and majority of projects there have had well over a year of just autocomplete before agents got their dirty hands on them anyway 😬

      • curbstickle_lw@lemmy.worldM
        link
        fedilink
        English
        arrow-up
        4
        arrow-down
        1
        ·
        edit-2
        4 hours ago

        Hey just to get this right, what you are asking is for me to edit the title to add [AIP]?

        Correct

        Regarding the disclosure, it is there

        Its not, please take a look at the examples. If you have questions, ask, but please don’t just say “its there”. There is a specific format being requested for a reason - it shows how you used AI.

        I can’t stress it enough I wouldn’t be able to tell you what’s AI and what isn’t

        You absolutely can identify what you need to put in, just take a look at the examples please and message with questions. Its not about “which line” its “here’s how it was used”.

        Sorry, didn’t realise it became THIS strict, won’t this just make people lie about it?

        A few have tried, yes, and got caught and removed.

        This was a decision that the community came to, as there was a non-stop flood of purely generated projects being posted.

        I can guarantee, given this is my work field, that as of today it’s gonna be near impossible to find any - good - software engineer not using AI, it kinda makes the flag redundant because it’s either you add it or you’re likely lying

        Its my field too, and again, any software engineer absolutely should be able to describe how they used it. If they can’t - and I’m not trying to be snarky here but direct - they probably shouldn’t call themselves software engineers.

        If you reference an external library, you can describe why you referenced that library, for x and y functions. This is no different.


        The [AIP] tag allows the anti-ai folks to filter out the posts they don’t want to see. Comments they typically make aren’t permitted, and get removed.

        Then you have folks that are OK with AI, but are more discerning in its use. Using me as an example, generically speaking I prefer no more LLM use than pair across the implementation, documentation more may be fine, design I want a person doing, so maybe hint at most, deployment I don’t care automate away.

        This is where https://ai-declaration.md/ came in. Which, BTW, you can include and automate, then just link here.

        Its a bit on the more complicated side, so that level of detail isn’t needed, and it was simplified some for here and the overwhelming majority of folks here were in full support of that as the approach.

        From your description, it seems like “assist” - AI acted on parts of tasks.

        Edit: and for the record, your post was reported for rules 7/8, explicitly for these reasons.

        • fccview@lemmy.worldOP
          link
          fedilink
          English
          arrow-up
          4
          arrow-down
          2
          ·
          3 hours ago

          Right, I hope I’ve done it properly, I think it should cover it all ♥️ sorry for not re-reading the rules, that’s absolutely my bad, I’ll make sure to continue with the disclaimers going forward 🙌

          Its my field too, and again, any software engineer absolutely should be able to describe how they used it. If they can’t - and I’m not trying to be snarky here but direct - they probably shouldn’t call themselves software engineers.

          My point was more that as a software engineer you’d be doing yourself a disservice by not using it, even then, I wholeheartedly disagree with that line. Maybe it’s the case if this is a small project you handle by yourself, but on large scale projects with more than one contributors there’s no way in hell you can keep track of AI usage without asking every PR if somebody has used AI, if something is done the right way it’s just code well written in my eyes, so I don’t even bother asking.

          And I know HOW it’s used as far as my workflow is concerned, btw, but it really does not feel that’s what’s being asked here, what I feel is been asked here is to what extent… this is not a library, come on… and if new features have been AI assisted does it mean the label applies to old features that were hand coded? It’s a confusing system, sorry, it only works for projects that have been agentic from the get go. Not to mention “assisted” or “pair” gives out the idea that a few prompts can make a project to this scale, which is absolutely not the case.

          Anyhow, just venting a bit, hopefully you won’t take any of this as an attack, I know moderating is hard work, and one thing we can definitely agree to is that the rules should apply to everyone regardless of opinions ♥️

          • curbstickle_lw@lemmy.worldM
            link
            fedilink
            English
            arrow-up
            2
            arrow-down
            1
            ·
            2 hours ago

            on large scale projects with more than one contributors there’s no way in hell you can keep track of AI usage without asking every PR if somebody has used AI, if something is done the right way it’s just code well written in my eyes, so I don’t even bother asking.

            You have it required for the PR, and contributors you trust you can take what they say as the case, new ones you review first. I do this all the time, even for internal datasets used to make models for embedded systems, saying its impossible is… Not true.

            Whether the code is the same is irrelevant, because its about understanding. Someone contributing code they don’t understand makes it functionally unusable - they won’t be capable of maintaining it. Something small enough that it wouldn’t matter doesn’t impact anything enough to even take it into consideration.

            Not to mention “assisted” or “pair” gives out the idea that a few prompts can make a project to this scale, which is absolutely not the case.

            They absolutely can with a large enough context. It will also have a ton of problems, but can be feature complete. There is someone I’m dealing with now who built a much, much larger system, entirely from prompt and generated code. To be candid I’m not even sure he reviewed it, which is why I’m now involved.

            That doesn’t make it - IMO - a good or maybe even useful project though. Purely generated can absolutely be large scale, but its maintainability suffers substantially. LLMs are going to give you the average response, not the best response, of everything ingested. Specificity in the prompt can tailor what that average looks like, but its still essentially an averaged response.

            Anyone is welcome to make a meta post about any rules btw, including these (and even me as mod - I promise I won’t be offended). They came from community meta discussions in the first place, followed by vote threads. I play more of a janitorial role than community dictator, I’m just enforcing the rules the community decides on.

            • fccview@lemmy.worldOP
              link
              fedilink
              English
              arrow-up
              2
              ·
              2 hours ago

              I play more of a janitorial role than community dictator, I’m just enforcing the rules the community decides on.

              Yeah I got that, which is why I wrote that last piece

              They absolutely can with a large enough context. It will also have a ton of problems, but can be feature complete. There is someone I’m dealing with now who built a much, much larger system, entirely from prompt and generated code. To be candid I’m not even sure he reviewed it, which is why I’m now involved.

              I’m obviously talking about the scale of the project as in continuous improvement, quality, standards, size, heck even age… All of it.

              What I posted cannot exist at the quality it exists by just prompting AI mindlessly without understanding what it’s generating. I’m sorry, I’m never gonna buy that.

              and even me as mod - I promise I won’t be offended

              I’d say you are doing your job pretty damn well and with a rare professionality, given how patiently you are replying to me, rather than just mindlessly enforcing your view, all good ♥️

              • curbstickle_lw@lemmy.worldM
                link
                fedilink
                English
                arrow-up
                1
                ·
                edit-2
                1 hour ago

                I’m obviously talking about the scale of the project as in continuous improvement, quality, standards, size, heck even age… All of it.

                What I posted cannot exist at the quality it exists by just prompting AI mindlessly without understanding what it’s generating. I’m sorry, I’m never gonna buy that.

                Sure, but thats not the part the average user is picking up on to be clear. Or even be aware of how it all works - a pretty hefty portion of self hosters are not programmers, or just getting into it, so that isn’t an aspect that comes across, it requires experience. I’m more referring to appearance of scale, as in large and complex.

                Further for those who do, there are different levels of individual comfort with LLM use. Some are OK with docs or test plans or dist pipelines being fully agentically managed, but don’t want to see a single line of code. Thats why the (simplest version I could come up with at least) disclosure helps there, individual decision making on the right fit for them.

                I’m a strong supporter of the disclosure I linked before outside of here, which is much more complex, even for internal tools being worked on at my employer (and clients). It makes developing a lot easier, too, to be candid. Makes bug hunting easier when I can expect a prompt over a person, or even a person over a prompt, because the types of mistakes made are often very different. Knowing a person wrote it or an agent wrote it makes fixes faster as a result. You may want to consider asking for LLM use (where/how and the prompt ideally) with PRs if only for that benefit.

                I’d say you are doing your job pretty damn well and with a rare professionality, given how patiently you are replying to me, rather than just mindlessly enforcing your view, all good ♥️

                Much appreciated! Landed here as mod in probably the most backwards way, but ive done this before on much bigger scales and I don’t get offended easily - too many other things in the world that are way more important to be concerned about after all.

  • qub@lemmy.ml
    link
    fedilink
    English
    arrow-up
    4
    ·
    2 hours ago

    Thanks for doing this! Been trying to self host stuff ever since I found out Google had betrayed my trust by using my private email and photos to train their models all these years. Work like this gives power to us who feel powerless in the face of Big Tech.

  • LsnLDN@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    14
    ·
    13 hours ago

    Oh my gosh you HERO I always thought open source software coders were just somewhere up in the clouds watching down on us but you’re actually addressing us mortals and now I do not feel worthy. Nah but seriously thank you so much for showing how the world can be something more than corpo greed, people like you are the last bit of hope I have in technology and maybe humanity!?

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      4
      ·
      11 hours ago

      Lmfao it does come across that way a little does it, dunno what I was thinking 🥲🤣 thank you for ending it in a good note btw ♥️

      • LsnLDN@sh.itjust.works
        link
        fedilink
        English
        arrow-up
        2
        ·
        8 hours ago

        I mean it all sincerely though 😂 everything in my life has been enshittified apart from open source software, thank you for your service o7

  • H Ramus@piefed.social
    link
    fedilink
    English
    arrow-up
    10
    ·
    17 hours ago

    Been using degoog since early days and watched it grow into what it is today. I’m not super technical and managed to get it working quite easily. Thanks for making this software!

  • Flatearth_Habeebi@lemmy.zip
    link
    fedilink
    English
    arrow-up
    13
    arrow-down
    4
    ·
    edit-2
    17 hours ago

    A note on the AI part. People think “anyone can vibecode anything”. Actually no, it’s not that simple. Sure, anyone can generate code with AI, but building something reliable and maintaining it is a different story. You still need to understand code, architecture, design and everything else. That way you can actually discuss work and features with AI. Make informed decisions. Find bugs and fix them. Find regressions and many, many other things that experience makes a huge difference in knowing how to deal with.

    AI needs so much guidance, and without the right knowledge and experience, you can easily end up with a shit product, if it even works at all. People need to realize this. It’s a very useful tool if you know how to use it and guide it right. It has helped so many projects catch up on technical debt. It has helped developers find flaws and vulnerabilities in their work.

    So many projects that I have been using for years have now been pushing updates much more frequently than they used to, and their projects have improved remarkably. Of course, AI isn’t without its problems, and there are real concerns about code quality, security and maintainability. But those things should be judged by the actual quality of the work, not just by whether AI was involved or not.

    This is very cool, and thank you for all the hard work and for sharing it. Much appreciated.

    I find it very disheartening and even disrespectful that some people can look at all the hard work developers put into projects they share with the world for absolutely free, and all the good they do for communities around the world, then just set all of that aside and focus on one thing they personally disapprove of. There’s nothing wrong with constructive criticism, but dismissing someone’s entire contribution because you don’t like one aspect of how they accomplished it is just unfair, man.

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      4
      arrow-down
      1
      ·
      11 hours ago

      This is very well said, it’s just sad that it has to be said to begin with 😆 plus if anyone thinks a 20$ Claude subscription can maintain that many huge projects on its own they are mad… shit’s expensive and I’m not about to go broke to maintain oss lol

      • Flatearth_Habeebi@lemmy.zip
        link
        fedilink
        English
        arrow-up
        2
        ·
        4 hours ago

        That’s a great point. I know that from experience. I have the $20 chatgpt/codex, and you can’t really code much with it. It is great for small features and bug fixes, but not really long hauls. And yes, it is very sad and really bothers me when I see it, especially on projects that you know for a fact have done actual hard work.

  • hanke@feddit.nu
    link
    fedilink
    English
    arrow-up
    28
    ·
    23 hours ago

    I use degoog as my main search engine since about a year back maybe, and I really like it!

    Works like a charm and pretty well decluttered.

    I appreciate you creating this software and releasing it for free like this!

    Just curious; Are you still considering this?

    https://github.com/degoog-org/degoog/issues/272

    I created the ticket because I thought that would be a great feature, but you are of course free to do as you please with it 😄

    Again, thank you for building and maintaining this! 🙏❤️

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      13
      ·
      21 hours ago

      Hey! I did not forget about that, I just parked it for now as I’m thinking of ways to crawl/index and eventually rank and if I am to re-make the entire ranking system it would be redundant to do it twice ♥️ everything that’s currently open in the issues is still there because it’s at least being considered :) thank you for using degoog ♥️

  • tuckerm@feddit.online
    link
    fedilink
    English
    arrow-up
    16
    ·
    22 hours ago

    This is amazing, I had no idea that the same person made Jotty and Degoog. Thank you so much!

  • sma3in@lemmy.world
    link
    fedilink
    English
    arrow-up
    6
    ·
    21 hours ago

    Thank for your work and contribution to open source self hosted software. I been using Degoog for the past 6 months and I really like it. I switch between searxng and degoog but for the most past, Degoog has been useful, and community plugins make it even more amazing.

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      11 hours ago

      Thank you! 4get is another amazing search scraper if either degoog/searx get blocked on your local instances 👀♥️

  • hellmo_luciferrari@lemmy.zip
    link
    fedilink
    English
    arrow-up
    6
    ·
    22 hours ago

    Thank you and all of the contributors hard work!

    I used to use SearXNG, but ran into many issues that I stopped wanting to futz with. I ended up trying a few other alternatives, but landed on Degoog and stuck every since. Its one if my most used self hosted apps.

    Thank you again!

    And as far as Scatola Magical goes, that sounds pretty neat! It would be cool to test it out!

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      5
      ·
      21 hours ago

      Hey! Thank YOU!

      Feel free to give scatola a try, it probably has a few mainly visual bugs and it’s a bit behind on third party dependencies needing update (I’ll sort it out this week) but otherwise it’s pretty functional! 👀

      • hellmo_luciferrari@lemmy.zip
        link
        fedilink
        English
        arrow-up
        3
        ·
        20 hours ago

        Ie thr best way for me to contribute via GitHub? Likely more bug report type than any actually contribution from me. But hey, if there are ways I can help, glad to!

        • fccview@lemmy.worldOP
          link
          fedilink
          English
          arrow-up
          3
          ·
          11 hours ago

          Bug reports and feature ideas are just as valuable as coding contributions as long as they are properly scoped and defined ♥️ so yes please!

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      11 hours ago

      You know, this comes out a lot, and the reality is that it would just be hugely out of scope, I think it’d best to have somebody make a systemd timer tool, which would probably be much better architected given it’d start straight with that in mind :)

      Not saying 100% no, but I have definitely considered it and decided against it a few times since I built it ♥️

  • Venia Silente@lemmy.dbzer0.com
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    10
    ·
    18 hours ago

    You say you use AI responsibly and not vibecode, but then at the next step you say you don’t know how much of the code is AI? Honestly sounds like a red flag. History is full of “cool old guys”, mentors and the like who took a nosedive into shit later on, so “15 years of history in FOSS” is not the boast and shielding on AI subjects you think it is.

    • fccview@lemmy.worldOP
      link
      fedilink
      English
      arrow-up
      9
      arrow-down
      1
      ·
      edit-2
      11 hours ago

      Oh yes, every engineer tracks exactly how many lines of code they write in their project, that’s realistic and to be expected 👍 and I can totally track if contributors use AI and it matters more than if the code is well written and the architecture is good 😬 also it’s 15 years of commercial experience, not 15 years of history in foss.

      And lastly I did not nosedive into anything, it took me a long time to “trust” llm outputs, a better contribution to this thread would have been asking what my approach, my workflow, my review style is and how long it took for me to even think if implementing agentic coding in my workflow :)