I think I found Jev is now available through OpenRouter? Oh but maybe you mean somethign like Bedrock with zero-data-retention offered? Anyway, it was on OpenRouter the other day, lest anyone be confused by above, as I did some testing with it for my pareto frontier tool https://github.com/bglusman/model_skyline (which, apologies, is full of slop because its 100% AI maintained but, may or may not have some utility for guiding automated or manual decisions... it definitely showed that Jev was performing MUCH better than some of the open alternatives to it we tested anyway)
Actually unless they’ve changed substantially Kimi kind of does this? It’s not as advertised/focused on and the API is a little nonstandard making it hard to plug some things into, but, unlike Claude they’ll actually give you an API key you can use directly as part of your unmetered subscription. I played with it for a month or two out of curiosity, it ls not a bad value though a bit brittle and stuff
This is a wild statement to me that “nobody seems to know about Gödel”… I think I learned about him when I was like 12? And while I’ve certainly met plenty of people who haven’t heard of him, until 10-15 years ago the same was true about Turing, mostly changed by a some Hollywood and/or literary references… and Gödel had a tiny bit of that in Oppenheimer if i recall, though not to the same degree? I bet if you asked 100 comp sci students if they’ve heard of Gödel at least 50 would say yes? On what basis do you think no one has heard of him? Not many have read the original math, sure, but… the idea that he was a mathematician who wrote an incompleteness theorem? VERY widely known.
Well… it doesn’t exist FROM APPLE or MICROSOFT or GOOGLE at their shipped OS Level, but… fundamentally this isn’t a “true OS” level feature you’re asking for, it’s something you think the OS products should bake in, and you might be right! But I think the parents post is suggesting YOU CAN BUILD a prototype of what you want, how it should work, on Linux…
I have a project somewhat close to this I’ve put on pause the last month or so, partly because I’m not sure how useful it is or where to take next, but I may incorporate Wayfinder into it as a next step to improve its capabilities, as part of what it is a model gateway/router that this feels like could make more powerful/flexible in its decision making. I can’t decide if what I’m building is mostly a model recipe cookbook/platform, or a debugging tool, or both or something else at the moment, but, it can do most of that… maybe it’s part of what you want, if you figure that out better? feedback welcome!
https://wardwright.dev/https://github.com/bglusman/wardwright
> Well… it doesn’t exist FROM APPLE or MICROSOFT or GOOGLE at their shipped OS
What I am saying does not exist period. What I am saying is that there isn't a proper abstraction that helps the ecosystem build upon it.
> But I think the parents post is suggesting YOU CAN BUILD a prototype of what you want, how it should work, on Linux
I mean, yes. But me saying "this does not exist" and someone saying "but you can build it" does not take away from the fact that... Yeah, it doesn't exit :).
And also, no, I cannot build it, at least not alone. Because I want apps to eventually build upon my abstractions. This would require a good set of millions, of which the technical development would be a small part. The coordination, contracts, API definitions, even marketing, etc would be the majority.
I am saying something that Google, Telefonica, Microsoft etc could do.
This is an interesting motivation for the project... I kind of get it, but, have you looked at fnox[0][1]? Curious how you'd compare/contrast goals with that if so, I think I prefer that as its not coupled to a single encryption tool (age) but supports age as well as multiple cloud or local options behind one unified interface... I think it can even mix multiple stores together? but I may be missing something/didn't read thoroughly yet...maybe there's a reason fnox doesn't work as well with Nix? fnox was discussed here previously[2]
I didn't know about it existence, it looks like a nice project! Also, it would probably play nicely with Nix (by writing wrappers instead of symlinks)
However it doesn't fit quite the same niche that my tool does. If I understand it correctly (though I only read those two links) fnox is more about how to use the secrets, ie given an encrypted description of the secrets, how to make them accessible to programs (kind of like sops?)
With my tool, secrets on the device are in plaintex and simply accessed by path reference by the respective programs. The focus of secs-man is more on exporting, ensuring integrity, and importing (possibly to remote machines). All of this, while being manually recoverable even without the tool.
Still, interesting project! Might take inspiration from it for some features
I think SRO's are very different from "One bedrooms" and even from studios? a studio or a one bedroom has a kitchen, a bathroom, and some living space... I think SRO's typically have a bed, a chair, and sometimes a desk. Bathrooms were shared, I'm not sure how kitchen/cooking space worked?
Curious for anyone who understands the science/optics here... if/when this is available, would it still have the downsides LASIK (and contact lenses) have for older people where you no longer need distance glasses, but now you may need reading glasses, when you didn't before, etc? Or might this be able to improve both distance vision and preserve nearsightedness/reading vision at the same time? That tradeoff is the reason I personally never got LASIK, as it was just trading one pair of glasses for another, for me...
There is no case where you need reading glasses and weren’t going to before lasik.
That’s a myth and completely unrelated.
You need reading glasses because muscles at your eyes lose the ability to pull tight and focus up close. LASIK is a correcting of the lens. Entirely different systems.
If you put off lasik because you thought you would need readers, congrats, you could have been seeing sharp at distance this whole time and will likely still need reading glasses.
There are several errors here. When the eye ages, the lens stiffens, effectively decreasing the focal range. For nearsightdness, Lasik alters the cornea and moves the relaxed focal distance outward. It's possible that a nearsighted person wouldn't need reading glasses even as their focal range decreases, but would after Lasik.
You can stand by things you said but also learn from them/from people’s responses to them…. For instance, you declare someone’s response virtue signaling… This hit me in a funny way, partly because it’s valid, it’s true, there is a lot of signaling that goes on you learn to see, virtue and otherwise… but also because of how insidious a criticism it is, because it reframes a debate away from correctness and towards who said it, whether they’re posturing…
I think it’s a category error and an ad hominem attack to bring it up in a debate with someone. It doesn’t mean your wrong or can’t still beleive they were virtue signaling, if that’s what you mean by standing by what you said, but more than one thing can be true and that being your reaction is not honest engagement with the criticism… I don’t care think it’s about the joke very much, it’s not especially funny but not all humor has to be, and I don’t love their reaction to it either, but I think you’re confusing the feedback you’re getting here and there and probably elsewhere that your opinions should change… a sibling comment spoke of being right vs effective, and there’s something to that, but there’s also being right vs having a growth mindset, about being open to genuine conflict that sometimes brings new perspective or insight… But that doesn’t happen when one side shuts down the other with ad hominem attacks or uncharitable assumptions. To be fair, it doesn’t happen online in mailing lists or discussion forums at all very often. Maybe you only get these kinds of reactions here and when people seem more real to you in person you engage differently… I know most people engage differently online than in person, and different pseudonymously than using real names. Someone else here compared you to Linus, and there’s probably something there? There’s no doubt you brought some vision and insight to both these projects, as he did, but something changed for him some years back that was a growth moment and caused him new perspective on how he engaged with people online. The same could still happen for you, and it wouldn’t mean you were giving in to a “woke mind virus”, it would mean you were growing.
I can’t respond to your response below but I fully agree “a lot of online criticism is not actually about truth-seeking or honest disagreement”, but I believe by ignoring the principle of charity, you undercut your own credibility and value. You may be able to show people how and where they’re in the wrong by demonstrating how THEY’VE made motive and framing the entire point, WITHOUT personally ascribing that as necessarily being a character weakness or hypocrisy or unconcern for the truth, but perhaps just a error on their part as well all make sometimes… just my $0.02
I was in the midst of obviously baseless allegations being made against me, not because of anything I actually said but because some very nasty[1] people disagreed with a naming decision I had made.
If you ever find yourself in that situation you are way past the principle of charity.
I'm not saying I couldn't have handled it more gracefully and probably would today, remember this was an obscure mailing list post from 3 years ago that someone dug up.
[1] This is not to suggest that everyone who disagreed with my decision behaved badly, it was a small minority
I just want to quickly jump on what you said about Linus. I know a lot of people look at his change and see it as a "growth moment", but my view is that he was forced to change by a growing body of people who take relatively extreme actions against those not seen to be towing the line. There was another group of people like this in history. We rightly condemned that evil group and their actions, and we were once more tolerant and open-minded towards one-another as people. I miss those days.
I think it starts with social coercion, intimidation, exclusion, economic pressure and ostracism long before it builds into the confidence to take more overtly violent measures. I don't know him, but it certainly appeared as though he succumbed to these pressures, given the timing of things. I hope people doing these things take some time to reflect on their actions and how closely they follow a dark path we've seen before.
Funny timing. I’ve been building something adjacent, though from a different angle: not primarily local-model reliability, but a control layer around agent execution, tools, routing, and operator intent. I was calling these "synthetic models", but decided yesterday "LLM middleware" is a clearer description.
The common thread I see is treating the harness around the model as first-class infrastructure. Forge seems focused on tool-call correctness and recovery; Wardwright is more about controlling what the agent is supposed to do, where work gets routed, and how the operator stays in the loop.
Curious whether you see those as complementary layers. I’m planning to try Forge and would be interested in seeing whether they fit together cleanly.
I've just read through your readme and I have zero clue what this does. Something about proxying model calls and applying "policies" to them? But what kind of things does it actually do, what benefits are there? That should be at the top of the readme.
I'm sorry to hear that! I'll take a fresh look at docs in my upcoming release.
In a nutshell, it applies guardrails around LLM calls to make them more reliable - specifically small models but works on all: "on multi-step agentic workflows through guardrails (rescue parsing, retry nudges, step enforcement) and context management (VRAM-aware budgets, tiered compaction).".
It'll try to parse malformed tool calls, it'll automatically compact if needed, it'll enforce any workflow requirements you define (ie, read before edit) - and it does so with domain-agnostic guardrails. It catches and feeds errors back to the model in a structured way so the model self-corrects (hopefully).
Each guardrail can be removed as desired by a consumer. It can be used as a building block library (WorkflowRunner approach), it can be integrated into existing source (middleware), or it can be a drop-in addition to an exiting workflow (proxy mode).
I think that comment was aimed at my Wardwright link, not Forge, given mention of policies and proxying model calls! I think your docs are in much better shape ;-)
Name was just a portmanteau of Calcifer's forge, because Howl’s moving castle seemed like a good metaphor for what I was trying to do… I had synthetic models as apiece there but I realized a) it was out of place and b) it was my favorite feature there
I’m not buying hosting from a password manager, I’m buying security. I don’t have complete confidence that I can secure a self-hosted password manager and it’s not an area where I want to take risks.
It's very simple, just don't make it accessible outside your home network. Clients sync when the server is accessible and use last synced data otherwise.
The effort required to set this up far outweighs the price to pay someone to do it for me.
I pay a cleaner, I have a dishwasher, I pay someone to do my taxes, I pay for companies to host software.
Then again, I never order food and almost never get takeaway, as cooking is nice and I value my food enough to care what goes in it. Cheaper too, easily offsetting what I pay for my password manager.
Tailscale for your laptop, phone, etc. to be able to talk to the other computers when away from your home WiFi. (Optional, but makes syncing easier).
Syncthing, talking to your Tailscale IP addresses if you use it, or your private WiFi network addresses if you don't use Tailscale.
One folder synced, containing keyfile2.kdbx.
30 minutes to set up and then you almost never need to think about it again. If you don't trust Tailscale, you can run a Headscale server or just not use it. And the syncing is entirely run on your machines; your data never ends up written to someone else's SSD.
I mean does it? I have set it up before but I just set it up for my new small office team. I already had an internal server and WireGuard vpn in our office and it took 2 minutes to create a quadlet to run vaultwarden and a few more to configure it. The “hardest” part was training the team on how to use collections.
reply