deleted by creator

If it worked for you, yeah. “If an idea is stupid and it works, it’s not stupid”, kind of sentiment (☞ﾟヮﾟ)☞

I’m glad to hear it! If you have time, it’s considered good manners to post the solution to your question if you find it, so anyone else with the same problem who finds the thread also finds the solution ☺️

The project I’m actually most excited about is exactly the “use the carpentry” one you pointed at: software for CNC and laser machines. I’ve got a laser at home that’s made me a lot of inlays and gifts over the years, and the existing tools (LightBurn etc.) are good but kept missing the technical, specialist features I wanted - so I started building my own. It’s called Nexus Studio.

Ayyyy yeah that’s what I’m talking about! (☞ﾟヮﾟ)☞

That’s a really cool project dude, if it’s public I’d love a link to check it out! (I tried Googling, to no avail) I have a friend who has both a CNC machine and a laser cutter (a fuckoff-huge thing the size of a desk, that can cut metal), so your work on that might be helpful to us 😊

In any case, good luck and I hope those resources help. Drop me a DM if you want to talk shop!

Any time, I hope they’re helpful! (☞ﾟヮﾟ)☞

I’m a little surprised to hear you say PAYG for Opus sub agents is economical

I did say it was surprising! 😂 To give you an idea what I mean by “economical”, it’s never more than a few bucks a day, even on days of heavy use and development with “loop until clean” instructions on QA (for which I use Opus). I accidentally blew through my opencode go quota really early in the first month, so I ended up on PAYG; here’s the usage graph:

And here’s the numbers breakdown for the highest day (I was evaluating GLM5.1 for general tasks - don’t use it for that, it’s really token hungry)

That includes a lot of experimentation too while I figured out which models were best for what. I hid Fable because it crushed the rest of the table - really expensive, but worth it for one-shotting very long tasks on the Anthropic subscription is what I found.

Ah, you’re alright mate. Like I said, I got the same guff when I posted my first project here, and it sucked. I wanted to offer my advice because I think it’s brilliant that AI is opening up development to a whole new raft of people, and I’m excited to see more people building neat little software projects that fill their needs - we get to watch peoples’ journeys unfold through repo commits. It’s like watching someone’s first Dark Souls playthrough.

I have another bit of feedback which I hope might be helpful: Everyone is building an encrypted messaging app these days. Do something new, and use your carpentry experience to inform what that something is. You have a wealth of knowledge and this thing is an incredible tool for augmenting that, see if you can solve an old hard problem with it!

See, now I want to help again 😂 What tools are you using and do you have any harnesses set up and what’s your system prompt like? The beautiful thing about agents is that you can just point them at repos and have them cherrypick the useful bits. Here’s a few that you may find useful:

• https://github.com/obra/superpowers
• https://github.com/DietrichGebert/ponytail
• https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f (strong recommend for organisation and human readability, also reduces token use handily)
• https://github.com/obelisk-complex/hermes-agent (this is my fork of Hermes that adds an on_output hook, but it includes my custom self-checking harness that I built for myself as well)
• https://github.com/obelisk-complex/claude-agents (My Claude Code agents, also good for cherrypicking generally but especially the QA agents)

I’m surprised you think I care. I’ve seen what you’re dumb enough to fall for.

The harness helps a lot even with local models. In fact, I just found this this morning and cherrypicked it: https://github.com/DietrichGebert/ponytail

Recommend doing the same, and for superpowers if you don’t have 'em already: https://github.com/obra/superpowers

Opencode Go is the $10/month cloud model subscription from the same group maintaining the OpenCode software. Opencode Zen is a pay-as-you-go version which gives you access to Claude models as well. Keeping pay-as-you-go to subagents only (e.g. telling your agent to launch an opus subagent via your opencode zen key) is actually surprisingly economical - when you’re not going turn after turn with hundreds of thousands of tokens of context, claude is pretty reasonably priced.

What I’m doing is spreading out my usage over multiple cheap subscriptions, and augmenting with the occasional pay-as-you-go frontier agent, to get quality in line with what you get out of Claude, at usage that would require the $200/month level, for a lot less money than that.

Imagine thinking you know all there is to know about AI and then using an AI Detector which are famously bullshit: https://mitsloanedtech.mit.edu/ai/teach/ai-detectors-dont-work/

You don’t need an AI detector. The patterns are obvious. But that’s also the wrong criticism. Rolling your own encrypted messenger app is dumb, whether you use AI or not. Setting up and hardening a Matrix server with AI == godbrain. That’s just implementing (battle-tested) software that someone who knows what they’re doing wrote.

I considered it, and the point stands. I came here offering advice - good advice, grounded in two decades of IT career, because nobody who cares about security rolls their own app with encryption unless they know what they’re doing. There’s too much risk of a bad implementation and leaving holes for bad actors to find.

They can just do what I do and use AI to set up their Matrix server. I set it up before AI was a thing too, but it’s so much faster now. That uses a lot less tokens, too. But they don’t seem particularly interested in actually taking advice onboard, so I’m not holding my breath.

edit: well that’s refreshing, he listened! Don’t get that on the internet too often these days.

You asked for criticism and advice, and that’s exactly what I offered. It’s based on my own experience here. I was offering it to be helpful. Just what exactly are you trying to achieve with this response? Because yeah, I’m not getting attacked, I’m getting lectured. For offering advice when it was solicited.

I’m certain you don’t see the problem here, so I’m out. But for the love of Christ, just use AI to set up, harden, and manage a Matrix server instead of wasting tokens building your own application. You’ll end up with something much more secure than an app with a possibly-bad crypto implementation that you don’t have the experience to see, find, or fix.

I’ve been using it with Opencode Go, Ollama, and Claude Code (it can delegate tasks to models through all those, so you can have Claude plan and Deepseek Flash build); I really like it.

I ran into that problem with the agent reporting that subagents succeeded, or work had been done, where it hadn’t (“I said I tested that, but I didn’t. That’s on me. Won’t happen again”), so I built a self-check enforcement system for it. You or your agent can set up the system by reading this: https://github.com/obelisk-complex/hermes-agent/blob/main/self-check-enforcement-system-v15.md

It includes the source patch which adds a hook on_output; this allows you to intercept text sent directly from the LLM to the user, which in vanilla is unblockable. So, this system ensures that if something remains unfinished, the LLM can’t say it’s done; it has to acknowledge what it didn’t do before it can send you a message to close the conversation loop. I’ve built the fork to automatically merge upstream changes around this patch daily at 0400 Pacific time, so I should stay up to date (ish).

I also put in a feature request to get this added upstream. Feature request here: https://github.com/NousResearch/hermes-agent/issues/45881

I think you’ll find I didn’t accuse you of trying to deceive anyone, I just suggested that you not gloss over using AI in your projects. Because yes, in my experience people here take it very badly if it looks like you’re trying to present AI generated work as your own (note: that doesn’t have to be your intent, it just has to look like it was). I was trying to be helpful, and as far as I can see my wording and tone were mild, so I see no reason that my comment should have made you feel unwelcome. The more fool me for the attempt, I guess.

Man, your writing here all reads like Claude after I’ve given it a list of “AI tells” to avoid in its writing. There’s structural patterns that are pretty easy to see when there are so many samples right next to each other. I strongly suggest not trying to gloss over your use of AI in your projects when posting about them; some people will always hate, but most I think don’t mind AI code as long as it’s been tested properly and doesn’t have any more bugs than you’d find in any other project.

Problem is, testing encryption properly is difficult, and there’s a lot more to a messenger application than just sending encrypted messages. That’s my criticism: you’re reinventing the wheel for no good reason.

My best advice is to set up a Matrix server if you really don’t trust Signal, rather than trying to roll your own. Its a lot less work, a lot more secure, and you can modify the source anyway if it doesn’t do what you need.

I typically prefer subbed; watching dubbed feels off somehow. Then again, I’m perfectly fine with the dubbing in DBZA so it must be a preference I learned somewhere!

What device are you watching on, and is it happening to all videos equally or does an MKV lag more than an MP4 for example? Are there any errors showing in the jellyfin logs?

Particles of light

Only half-right.

If you want the FOSS community to be there for you, don’t go out of your way to use the corporate BS that’s actively destroying it.

Heheh, and another one of you thought to call me hyperbolic.

intentionally designed to cause disruption to a computer, server, client, or computer network

Actively and deliberately adding text with the explicit purpose of attempting to delete other peoples’ work fits this quite nicely, thank you for including it to illustrate my point.