I've been hacking away at a browser-based tool that uses anthropic APIs on the backend. But what I really want is for the browser to talk to my local claude becuase I have MCPs, skills, network access for a bunch of things.
I started with a little proxy installed on my computer that the browser can call but knew it would never pass any security review. The alternative I didn't originally know about was Native Messaging.
It's a fairly benign way to let a browser talk to and execute commands on your computer. But doing it without disclosing is, I agree, very bad.
Nothing wrong about running http server on your localhost and talk to it. A lot of applications do that. The best thing: you don't need to appease extension appstores, you just ship.
The only nuance is that recent chrome versions treat it as a separate permission, so user need to allow it once.
Yes, native messaging is the "proper" way to do that, but, again, nothing wrong with localhost http server. You have origin headers so you can allow access from your whitelisted website, if necessary.
You only have origin headers that you can trust if the traffic originated from a browser you trust.
Anything else on the machine that can send network traffic can now hook into your service. Which is quite a bit looser than being able to start a new process running that native message host and hook into its stdio.
This has been great for me too. Every feature starts and completes with /gh-issue <issue number>.
Every issue is created with /spec and a conversation with a human. Once the spec is materialized as an issue it’s sufficient for an agent to implement.
I've been thinking the opposite. It sucks to be in the generation of workers that are displaced by AI. It's going to be great to be in the generation where work just isn't something that humans are expected to do.
That's what the whole UBI thing was about though. People did see this coming and wanted to preempt it. I'm not sure whether it would've worked, but people did try to come up with solutions for this transition period.
We are never going to live in a society that doesn't expect people to work. There may not be enough work for half the population, but people will still be expected to work to live. We already live in a society that could feed every last poor person and we still choose not to, cuz "but muh tax dollars!"
I mean, assuming we don't hit some limit with AI, we're going to get to the point where the best way humans can affect productivity is to just get out of the way.
Code quality and IDEs aren't going anywhere, especially in complex enterprise systems. AI has improved a lot, but we're still far from a "forget about code" world.
Yes it was there (not in all of course, but in some), in fact that is where the concept came from - it's necessary when maintaining large systems to keep the code consistent and clear.
I don't think we are. We will not be able to keep the peace with code production velocity and I anticipate that focus will be moved strongly to testing and validation
> code quality as we previously understood will not be a thing in 3-5 years
Idk - I feel like the exact same quality, maintainability, readability stuff that makes developers more effective at writing code manually also accelerates LLM driven development. It's just less immediately obvious that your codebase being a spaghetti mess is slowing down the LLM because you're not the one having to deal with it directly anymore.
LLMs also have the same tendency to just make the additive changes needed to build each feature - you need to prompt them to refactor first instead if it's going to be beneficial in the long run.
I've found that models have improved here significantly in past few months. They have the tendency to pile on ad-hoc solutions by default, but are capable of doing better architectural decisions too if asked.
A better design can be made somewhat default by AGENTS.md instructions, but they can still make a mess unless on a short leash.
We've seen a steady shift in music over the past 2 decades from full length albums, to single hits, to artificially generated.
Surely there's some gained and some lost. But coming from the era of buying an entire album, spending time reading the CD booklets and art, and listening to 10 songs which tell a larger story ---- what's being lost really hits home.
This comment is like 20 years out of date haha. People shifted to single hits when the iTunes store was selling songs for 99 cents. Now (and by now I mean for over a decade) we’re in the age of streaming, and you can easily access whole albums with zero friction. It’s the best time ever for the full listen through experience. And artists are responding by releasing long albums.
What I do think is lost these days is listening to the save album over and over again.
I really don't think we have. When I was growing up in the 90s it was the heyday of the pop single but there were still plenty of albums being produced and I think it's the same today.
I can tell you that myself (and many others) still create concept albums as our primary format. It's not that people aren't still creating it.
The choice is still there for any listener that cares about albums as a format. I don't mean that in a negative way. I suspect that many people listen to both playlists of singles, and albums of their favourite artists, depending on mood.
No, the game has changed. Back then, the singles were typically accompanied by an album, even if it was just filler. It's better to release singles now due to the way the Spotify and iTunes algos work. Best practice is now to release your songs one at a time rather than a full album (at least if you aren't an established player).
On one hand this pretty much destroys thematic albums (like classical music, prog rock, Tool or for example, something like Alice in Chains' Dirt), but on the other few could pull it off and those who can are still doing it (ex: the latest Opeth album). So maybe discovering new music is hurt, because itunes and spotify look like crowded ERs, but there's just as much good music out there - regardless of your tastes.
Right, there's less unnecessary dressing of an "album" of filler. But I don't think that's a meaningful change. Singles drove the market then and they do now. Albums were still produced then and still are now.
Right but the approach is different. Now the idea is as soon as you have a song, you record and release it. You don't wait until you have 45 minutes or 12 tracks or whatever. The album comes out later, but it's just a collection of songs (and that arguably isn't really new either).
ah, the standard trite, reductive anti-pop cudgel.
no, these days, pop albums are more frequently meant to be consumed in their entirety, often with full length visuals for each song that blend into each other in order.
* the death of radio has really meant that singles are declining in utility, especially in our social media era where the songs that pop off an album are not necessarily the record-designated singles
* the more parasocial development of pop encourages fans to invest more in merch and the concept of the album
* like everything else in the economy trending towards more expensive but meaningful experiences, tours are becoming larger productions to experience an album intensely
* in the AI era, we are now seeing artists pivot towards doubling down on experiences that AI cannot curate and provide meaning for
Rosalia this year is touring with a full orchestra and RAYE with a full big band, because these are intentional choices that the pop music industry has been trending towards for a while. There's always going to be trite drugstore music as long as there are drugstores, but what is charting is not really that at the moment.
I feel like in those days I really didn’t appreciate albums. Storage was a premium so I would focus on bands greatest hits songs vs discographies. Both in terms of my burned cd collections and early mp3. I didn’t start getting into albums until terabyte hard drives were cheaper. Then I started pirating discographies and listening to the back catalog for the first time.
music has been a product of its form factor for a long time. It's no coincidence that the wax cylinder, 78, 45, 33, cassette, CD, and mp3 dictated changes in how music was packaged (single, lp, ep, album, b sides) and the average length of a popular song.
Good thing music as a topic is diverse and people are doing all kinds of things. But yes, commercially distributed mass-consumption music is influenced by its packaging and distribution ... obviously.
There is another new trend now that some artists are doing album concerts, where the set list follows one of their albums. I thought that was cool.. probably as a reaction, on how to bring back the album in a way.
One can still buy artisan albums created by independent singers/bands. But they tend to get lost in the marketing/influencer noise and thus do not get worldwide success. As a result you have to search harder for them.
it's become a much larger section of music. Notice, there are no bands any more. Try finding a metal band with musicians under 40, or a Greenday, Linkin Park or, the itself automated Kpop industry aside, a spontaneous boy or girl group.
Solo pop or hip-hop performers with a focus on social media have crowded out collectively made music, likely due to the general social atrophy and technology enabling production from their bedrooms.
Anecdotally, I used to be a guitarist and a lot of my friends are musicians and teachers, teenage bands are pretty much nowhere to be seen.
Yup, the manager gets implicit credit for the work their team does. In most cases, deservedly so. I don't see why it should be any different for engineers using LLMs as "direct reports". Not all engineers will be the same level of "good" with LLM tools so the better you are (as with any other skill as well) the more credit you would receive.
This is basically the same workflow I've come to adopt. I don't use any "pre-built" skills, mine are actually still .md files in the .claude/command/ folder because that's when I started. The workflow is so good, I'm the bottleneck.
I've started to use git worktrees to parallelize my work. I spend so much time waiting...why not wait less on 2 things? This is not a solved problem in my setup. I have a hard time managing just two agents and keeping them isolated. But again, I'm the bottleneck. I think I could use 5 agents if my brain were smarter........or if the tools were better.
I am also a PM by day and I'm in Claude Code for PM work almost 90% of my day.
I like Claude, at least when the user reviews the code before asking for a PR. But gods I hate tickets/feature requests written by Opus/Sonnet (or worse: Codex or Gemini). If you know/understand your product enough it's probably less of a problem for your team than it is for mine, but each time I see a feature request automagically written in the backlog I know I will have to spend at least 30 minutes rewriting in so that it doesn't take us one hour to refine it collectively.
A bit, but mostly it propose extremely well-rounded solutions that are almost never complete, and sometimes miss a major point. I would rather have my juniors work themselves to understand what is needed, or/and ask me questions rather than follow the ticket that is basically a Claude plan. Right now I am modifying and object that was incomplete and I will have to do a migration because I didn't catch the missing attribute during the PR. It isn't big, and we could have coded workaround instead of redesigning the object, but: workarounds complexify the code, the data is less intuitive, and that also means the person who wrote the original object do not really understand the goals.
With a less 'expensive' ticket, with less explanation about how things should be done, but why they are needed, we would have had discussions, in dailies or 1on1, and that could have been ironed out then.
Yeah, basically Claude generate tickets that are heavy on the 'how' and light on the 'why', and I think that should be the other way around, for multiple reasons, but I'm already long-winded.
I buy it. SaaS doesn’t have to go extinct for this to be true.
I’m building an app and many things I’d normally pay for like metrics and emailing I can just do myself.
A friend has a law firm employing 100+ people and they are building so many internal tools they would otherwise be delaying or paying salesforce consultants for.
You really should pay, especially for work by small foundries.
Making a typeface takes a tremendous amount of work. The financial upside is extremely hard to justify.
I think non-designers underestimate the amount of effort required by an order of magnitude. I put it in the territory of building indie games. Potentially years of your life go into it, and it's a huge problem if everyone pirates your work.
I'm actually a designer, have paid for many fonts - including licenses for websites - have made a couple myself and have a good idea how hard they are to make.
That said, a certain corporation's bought up a load of fonts made over the past x decades and is making a tidy sum selling old rope again and again without adding anything of value, or funding the original designers/converters, so I'm quite happy to illuminate how an individual can get around such things for use on their personal blog with an audience of ten, should they so wish.
.
ed - you're also not as likely to be able to get a whole usable font from a small foundry in the first place, without buying it.
Fonts are (simplifying greatly) just code, right? I wonder when AI models will be able to cleanroom-clone the general look of a font without violating the copyright on the official version's underlying code.
I do a bit of graphic design for friends and family and am more than willing to spend like up to $100 on a nice typeface from smaller creators. It's just unfortunate many professional typefaces from the big foundries will charge you thousands of dollars for an entire family with strict usage limits. Like I'm just trying to make some holiday cards...
I get that they're trying to make their ROI back from enterprise customers who can justify paying thousands a month for their specific corporate font, but I would like to see more personal use, project based pricing that's affordable for hobbyist use
+1 — "just being lazy" is no excuse when you can just ask your LLM of choice for a free font recommendation similar to what you can't afford. If you absolutely can't live without using the paid font, of course you should pay for it!
You can't copyright basic geometric shapes either.
>there are plenty of perfectly legible typefaces that are completely free for you to use.
You mean there are plenty of bezier curved shapes which are within the public domain and no one can stop me. I'm not obligated to surrender my rights so you can turn typography into a business model. If you piss me off, I might just release tools that even stooges can use that copy the shapes out of the otf file, rearrange those completely so that no file fingerprinting will match, and has the user rename the files. I will go to war.
Legally, typeface designs do not receive protection (which is based on idiotic declarations like “you can’t copyright the alphabet”) but digital font files are considered programs and thus are able to be protected as IP.¹ You can try to justify the theft to yourself but somewhere there’s an individual (or on some occasions many individuals) who spent a long time making decisions about how that typeface should look and choosing the best points to turn it into splines to describe the shape and you decided that your laziness trumps their work.
⸻
1. I would note that bitmap fonts do not receive the same protection as Type 1 or OTF fonts.
>there are plenty of perfectly legible typefaces that are completely free for you to use.
Legally based off the carefully considered positions of philosophers of law like Thomas Jefferson, and others just as renowned, who actually created modern copyright law in the United States, because they weren't trying to set you up to be rent-seeking degenerate scribblers for the next umpteen millennia.
>but digital font files are considered programs
As precedented in case law by degenerate judges who should be brought up on treason charges. They aren't programs in any meaningful sense, culinary recipes are likely closer to programs (they, arguably, run on a Turing-complete machine, the human brain, and have something akin to branching going on once in awhile).
>You can try to justify the theft
What theft? No theft occurred, because I denied no one the possession of their own property. Even the judges and lawyers have to admit that this is at most infringement, so please use that word or just confess here and now that you'rea manipulative liar.
>I would note that bitmap fonts do not receive the same protection as Type 1 or OTF fonts.
What!?!?! Those aren't programs too? Please, consult the computer scientists, they must be informed! Are they also not stored as ones and zeroes?
I've been hacking away at a browser-based tool that uses anthropic APIs on the backend. But what I really want is for the browser to talk to my local claude becuase I have MCPs, skills, network access for a bunch of things.
I started with a little proxy installed on my computer that the browser can call but knew it would never pass any security review. The alternative I didn't originally know about was Native Messaging.
It's a fairly benign way to let a browser talk to and execute commands on your computer. But doing it without disclosing is, I agree, very bad.
(tool I'm hacking away at needs to talk to local claude and acli: https://withlattice.com)
reply