parasti

This is entirely based on the "agent skills" system. LLM agent only sees the one-line skill description in its context and "lazy loads" the rest of the skill file on demand.

Far better results compared to GPT 5.4 and Opus 4.6. Not great for execution due to speed but has consistently had better comprehension of the codebase. Maybe it's a case of "holding it wrong" regarding the other models but that's been my experience.

We're using Augment Code heavily on a "full rewrite of legacy CRM with 30 years of business rules/data" Laravel project with a team size of 4. Augment kind of became impossible to avoid once we realized the new guy is outpacing the rest of us while posessing almost no knowledge of code and working fully in the business requirements domain, extracting requirements from the customer and passing them to AI, which was encoding them in tests and implementing them in code.

I'm using `auggie` which is their CLI-based agentic tool. (They also have a VS Code integration - that became too slow and hung often the more I used it.) I don't use any prompting tricks, I just kind of steer the agent to the desired outcome by chatting to it, and switch models as needed (Sonnet 4.6 for speed and execution, GPT 5.1 for comprehension and planning).

My favorite recent interaction with Augment was to have one session write a small API and its specification within the old codebase, then have another session implement the API client entirely from the specification. As I discovered edge cases I had the first agent document them in the spec and the second agent read the updated spec and adjust the implementation. That worked much, much better than the usual ad hoc back and forth directly between me and one agent and also created a concise specification that can be tracked in the repo as documentation for humans and context for future agentic work.

Not in disagreement, but based on how casually the frontpage throws around terms like "X25519 keypair", it is obvious that mass adoption and ease of use aren't among the goals of this project. Looks more like an exploration of a concept - can a social network viably exist without any middlemen in between.

I would fault nginx here instead. Tilde-suffixed backups is an old convention used by vim and emacs among others. Definitely older than nginx.

Cool tip - never knew this. I always figured piping to `tee` is a must in order to view-and-save command output at the same time. Turns out I can do "command >&1 >file.txt" instead!

The biggest roadblock to using agents to maximum effectiveness like this is the chat interface. It's convenience as detriment and convenience as distraction. I've found myself repeatedly giving into that convenience only to realize that I have wasted an hour and need to start over because the agent is just obliviously circling the solution that I thought was fully obvious from the context I gave it. Clearly these tools are exceptional at transforming inputs into outputs and, counterintuitively, not as exceptional when the inputs are constantly interleaved with the outputs like they are in chat mode.

What does "vanilla C" mean? Asking as a maintainer of 20 years of a game written in C.

> We do not optimize for time spent in ChatGPT.

So ChatGPT constantly ending all responses with tangents and followups is not for engagement?

Been using git since 2008, and this looks more intimidating than helpful. git rebase takes a list of patches and applies them one by one on top of a given "base" commit. And that's it. It's not all this complicated git command soup. It's just patches. No objects, no sha1s, no metadata, no branches, just literal textual diffs applied in order. It's the dumbest and also one of the most powerful things about git if you care at all about a readable history. If you don't, that's fine, but in some circles, e.g. most (if not all) open source projects, patch management and history hygiene is a very important part of good collaboration.

The tech that you are defending is going to put your profession out of work. If you can write a prompt that gets you exactly what you have imagined for your movie, then your entire craft has been obsoleted and can now be achieved by anyone with an imagination (or even without).

You're romanticizing, sadly. Every time I see someone scratching off numbers, I see a twisted industry exploiting human hopefulness and naivety. Dreaming costs nothing.

I really tried to get into the vibe coding thing - just describe the thing I need in human language and let the agent figure it out. It was incredible at first. Then I realized that I am spending a lot of time writing clarifications because the agent either forgot or misinterpreted something. Then I realized that I am waiting an awful long time for each agent step to complete just to write another correction or clarification. Then I realized that this constant start-stop process is literally melting my brain and making me unable to do any real work myself. It's basically having the same effect as scrolling any other algorithmic feed. Now I am back to programming myself and only bouncing the boring bits off of ChatGPT.

Implicitly you appear to be saying that we need to reach that point before action is taken?

I upvoted but I was not taught this! I have had to slowly figure it out on my own. Writing things down is kind of like augmenting your brain. It's a memory that does not forget. When working through a problem, writing it down tends to point out the holes in your understanding. A corner case is never lost or forgotten when written down, it just stares at you until you write down a solution. The next step after realizing this is to develop the discipline to write things down and to organize your environment so it's effortless to write things down.

"Eating sugar" and "reasonable amounts of sugar in your diet" are two very different things.

I don't disagree but I feel like having spent a couple of decades filtering bad Linux advice on sketchy forums has honestly built up some resilience.

As a EU citizen, I don't care? People are not swayed by political ads. They're swayed by influencers who push political agendas covertly.

First-home buyers aren't the ones causing prices to go up.

Yeah, this is just an algorithmic feed - or, at least, the first attempts. After a while you'll stop asking questions yourself and just scroll the feed, just like you do on every other platform.

What's Microsoft thinking here? We got Windows on our kids' laptops and it's a pain to do anything with every time, just an amalgamation of decades worth of UIs held together with duct tape, looks terrible and performs even worse than it looks. When I'm thinking of the next big upgrade, Windows isn't even on the list of options anymore, and that's not even an ideological statement in any way.

I recently learned that Tiktok has a thing called "Streak Pets". Imagine taking a dopamine addiction-inducing activity and imagine gamifying that to maximize engagement in that activity. Imagine the brainstorming sessions at Tiktok where they navigate around the glaring issue of the fried brain circuitry of their own users.

Oops, one of the dependencies is a C++ library that doesn't compile on your less-than-five-years old arch.

First step is recognizing the problem. It'll now be easier to recognize when you've slipped into that mode and stop.

Second step is breaking the habit. Literally, you need to start spending that time on something else. Your behavioral autopilot automatically gravitates to browsing Youtube - you need to rewire that to something else. For some reason it's really difficult the first couple of times and then gets easier.

Third step, once you're starting to feel like you've made it out, is understanding that there are circumstances out of your control that will lead you back into those old habits. Maybe work was draining or relationship problems hit - whatever it is, you'll feel like you deserve a break. You need to recognize those moments and have a plan ready for those moments. You absolutely deserve that break, but falling back into that habit is literally self-sabotage, and you'll feel worse after.

Fourth step, rinse and repeat. I've never made it past this step.

I read "cancer" in between the lines of that comment. So the characterization of (potentially) that backdrop as "a bit of personal harm" feels wildly overassuming.

I feel there's space for brainstorming and creating new ways of making web apps without having to take a stand against the status quo. It's a fun thought exercise, naming all the things you think are wrong with the web, but I had to scroll real far to see that this is a post about Use.GPU: "Use.GPU is a set of declarative, reactive WebGPU legos. Compose live graphs, layouts, meshes and shaders, on the fly." So felt like a missed opportunity to me to highlight that more instead of going through the list of annoyances.

I am not arguing that.

You called a viewpoint that I share "utter nonsense". So I presented a counterpoint. A counterpoint to your opinion is neither a troll nor an attack on the original author's intent and/or character.

All I'm saying is we can both read the same article and one of us goes to build an app for their kid while the other goes and builds an app that gets other people's kids addicted to it while shoving ads. This has nothing to do with the author of the article.

As a parent of a toddler, could not disagree more.

This article can be read two different ways.

A parent making an app for their own children is wholesome and making a list of UX finds is helpful to other parents in that position.

At the same time app stores are filled with games that hook kids on with bright visuals, sound effects, and basic button mashing.

That second use case is vastly more common.

My boss posts GPT output as gospel in chats and task descriptions. So now instead of being a "you figure it out" it's "read this LLM generated garbage and then figure it out".

This guy ports open source games to mobile devices and sells them under their original names. (Selling is okay, misrepresenting them as the official version is not.)