wild_egg

8 ms·

91.
by wild_egg 4mo ago
No it doesn't.
92.
by wild_egg 4mo ago
I'm a divided individual. I've spent an obscene number of hours learning how to get reliably good quality code out of these things. I'm actually very happy with where the tech is right now and can't imagine ever going ba
93.
by wild_egg 4mo ago
Which is what makes putting an LLM inside a lisp so much fun
94.
by wild_egg 4mo ago
FilePilot has been a thing for a while now
95.
by wild_egg 4mo ago
> why are you comparing Claude Code to your editor? Because the editor does more. All the compute-intensive parts of the agent are in the cloud. Zero reason for an agent harness to require anything beyond a potato to run.
96.
by wild_egg 4mo ago
What does a breakthrough look like?
97.
by wild_egg 4mo ago
Deferred tool loading is not part of MCP. It's a Claude API special parameter that most other LLM APIs do not support.
98.
by wild_egg 5mo ago
There's a VS Code extension that got me through the Common Lisp learning curve using my familiar tools https://marketplace.visualstudio.com/items?itemName=rheller....
99.
by wild_egg 5mo ago
We train dogs to be subservient but that doesn't automatically mean we anthropomorphize them
100.
by wild_egg 5mo ago
For the real emacs experience you could use this mod to render an IDE in Minecraft editing the mod that renders the IDE.
101.
by wild_egg 5mo ago
When you say modules, is that @scope or something else? I can't find any reference to a native thing called modules but this seems to fill the same role.
102.
by wild_egg 5mo ago
That still leaves the question of how one gets their foot in the door. Lots of us are aware of the budgets but we don't get how's sales work at that level.
103.
by wild_egg 5mo ago
> $32 / 1M audio input tokens ($0.40 for cached input tokens) Anyone know how much audio is 1M tokens? I have no way of knowing if this is fine or prohibitively expensive.
104.
by wild_egg 5mo ago
Definitely not 50%. Aside from the aforementioned anticheat titles, my entire Steam library works flawlessly and has for several years now. I'm sure there's some long tail of titles with poor support, but how many people are actua
105.
by wild_egg 5mo ago
Unless your games require kernel anticheat, there's a decent chance they actually run better on Linux than Windows these days.
106.
by wild_egg 5mo ago
Can get a GUI with the same prompt if you tell it to use TCL/tk instead of Go + Charm stuff
107.
by wild_egg 5mo ago
Skill issue. Literally. Make a SKILL.md that has the agent leverage subagents to do all work. An implementor agent does the thing, and then a separate agent reviews and verifies afterwards. The fresh context window of the second agent doesn
108.
by wild_egg 5mo ago
Being able to fire someone, thus causing potentially significant hardship, is considered quite normal and acceptable in most lines of work.
109.
by wild_egg 5mo ago
Hey, so... Thanks for this. I've been building ticket systems and agents and whatever else as flat files in git repos lately and now I see I have to extend that to actually managing the repos themselves. This is going to be so nice.
110.
by wild_egg 5mo ago
If you know lisp, just reach for Coalton instead of Haskell
111.
by wild_egg 6mo ago
Technically, I think this is meant to develop Coalton, which is also statically typed and incredibly effective as a language for agents. All those ergonomic benefits that humans enjoy also allow AIs to develop lisp systems quite rapidly and
112.
by wild_egg 6mo ago
At the core, they're really very simple [1]. Run LLM API calls in a loop with some tools. From there, you can get much fancier with any aspect of it that interests you. Here's one in Bash [2] that is fully extensible at runtime th
113.
by wild_egg 6mo ago
Inference, in and of itself, can't be completely unprofitable. Unless you're purely talking about Anthropic? But > If you want LLMs to continue to be offered we have to get to a point where the providers are taking in more mone
114.
by wild_egg 6mo ago
A week of downtime every decade I think still works out to a higher uptime than I've been getting from parts of GitHub lately. So I'd consider that a win.
115.
by wild_egg 6mo ago
Where did you see a haiku comparison? Haiku 4.5 was my daily driver for a month or so before Opus 4.5 dropped and would be unreasonably happy if a local model can give me similar capability
116.
by wild_egg 6mo ago
That's really not remotely the same thing
117.
by wild_egg 6mo ago
Haven't the SQLite tests always been closed? Getting access to them is a major reason for financially supporting them
118.
by wild_egg 6mo ago
It only takes 20 minutes and $200 to hack a closed source one too though. LLMs are ludicrously good at using reverse engineering tools and having source available to inspect just makes it slightly more convenient.
119.
by wild_egg 6mo ago
That's exactly the message I got from the video
120.
by wild_egg 6mo ago
I hate the feeling of playing roulette with my account every time I use their tools. Since they refuse to actually provide definitive rules or policies, I have fully moved off their models and actively encourage all the other devs I know to

More ›