Hacker News: kixpanganiban

Great device, wrong problem: Two months with the Ultrahuman Air

kixpanganiban — Sun, 30 Nov 2025 03:13:38 +0000

Article URL: https://kix.dev/great-device-wrong-problem-two-months-with-the-ultrahuman-air/

Comments URL: https://news.ycombinator.com/item?id=46093272

Points: 2

# Comments: 0

First make it fast, then make it smart

kixpanganiban — Wed, 26 Nov 2025 08:14:56 +0000

Article URL: https://kix.dev/first-make-it-fast-then-make-it-smart/

Comments URL: https://news.ycombinator.com/item?id=46055272

Points: 3

# Comments: 0

Dumb Cursor is the best Cursor

kixpanganiban — Sat, 25 Oct 2025 07:19:34 +0000

Article URL: https://kix.dev/dumb-cursor-is-the-best-cursor

Comments URL: https://news.ycombinator.com/item?id=45701943

Points: 2

# Comments: 0

Unhogging RAM from WSL

kixpanganiban — Thu, 23 Oct 2025 09:18:48 +0000

Article URL: https://kix.dev/unhogging-ram-from-wsl/

Comments URL: https://news.ycombinator.com/item?id=45679874

Points: 4

# Comments: 0

New comment by kixpanganiban in "Two things LLM coding agents are still bad at"

kixpanganiban — Fri, 10 Oct 2025 04:53:53 +0000

> The article is confusing the architectural layers of AI coding agents. It's easy to add "cut/copy/paste" tools to the AI system if that shows improvement. This has nothing to do with LLM, it's in the layer on top.

I think we can't trivialize adding good cut/copy/paste tools though. It's not like we can just slap those tools on the topmost layer (ex, on Claude Code, Codex, or Roo) and it'll just work.

I think that a lot of reinforcement learning that LLM providers do on their coding models barely (if at all) steer towards that kind of tool use, so even if we implemented those tools on top of coding LLMs they probably would just splash and do nothing.

Adding cut/copy/paste probably requires a ton of very specific (and/or specialized) fine tuning with not a ton of data to train on -- think recordings of how humans use IDEs, keystrokes, commands issued, etc etc.

I'm guessing Cursor's Autocomplete model is the closest thing that can do something like this if they chose to, based on how they're training it.

New comment by kixpanganiban in "Two things LLM coding agents are still bad at"

kixpanganiban — Fri, 10 Oct 2025 04:50:17 +0000

In my defense, I wrote the blog post about quitting a good while after I've already quit cold turkey -- but you're spot on. :)

Especially when surrounded by people who swear LLMs can really be gamechanging on certain tasks, it's really hard to just keep doing things by hand (especially if you have the gut feeling that an LLM can probably do rote pretty well, based on past experience).

What kind of works for me now is what a colleague of mine calls "letting it write the leaf nodes in the code tree". So long as you take on the architecture, high level planning, schemas, and all the important bits that require thinking - chances are it can execute writing code successfully by following your idiot-proof blueprint. It's still a lot of toll and tedium, but perhaps still beats mechanical labor.

Two things LLM coding agents are still bad at

kixpanganiban — Thu, 09 Oct 2025 04:33:48 +0000

Article URL: https://kix.dev/two-things-llm-coding-agents-are-still-bad-at/

Comments URL: https://news.ycombinator.com/item?id=45523537

Points: 345

# Comments: 370

Python feels sucky to use now

kixpanganiban — Sat, 04 Oct 2025 09:03:37 +0000

Article URL: https://kix.dev/python-feels-sucky-to-use-now/

Comments URL: https://news.ycombinator.com/item?id=45471875

Points: 7

# Comments: 0

We should stop pretending like LLMs are software engineers

kixpanganiban — Sat, 27 Sep 2025 10:21:52 +0000

Article URL: https://kix.dev/we-should-stop-pretending-like-llms-are-software-engineers/

Comments URL: https://news.ycombinator.com/item?id=45394573

Points: 6

# Comments: 0

Cutting the cord on TV and movie subscriptions

kixpanganiban — Fri, 26 Sep 2025 09:25:04 +0000

Article URL: https://kix.dev/cutting-the-cord-on-tv-and-movie-subscriptions/

Comments URL: https://news.ycombinator.com/item?id=45384510

Points: 5

# Comments: 0

Complacency is the clear and present danger

kixpanganiban — Thu, 25 Sep 2025 09:55:39 +0000

Article URL: https://kix.dev/complacency-is-the-clear-and-present-danger/

Comments URL: https://news.ycombinator.com/item?id=45370992

Points: 3

# Comments: 0

Video: Is Lex Fridman a Fraud? [YouTube/Internet Anarchist] [video]

kixpanganiban — Sat, 20 Sep 2025 10:07:54 +0000

Article URL: https://www.youtube.com/watch?v=BiQoX01MS8U

Comments URL: https://news.ycombinator.com/item?id=45311968

Points: 4

# Comments: 0

New comment by kixpanganiban in "Munich from a Hamburger's perspective"

kixpanganiban — Fri, 20 Jun 2025 06:35:08 +0000

I clicked so fast expecting some sort of weird art experiment of maybe a hidden camera stuffed into a delicious cheese burger, as it goes through Munich.

Sadly, I was disappointed.

New comment by kixpanganiban in "Show HN: Doctor – tool to crawl and index websites and MCP server for LLM agents"

kixpanganiban — Tue, 13 May 2025 01:24:53 +0000

Hi! I wrote Doctor because I keep struggling with grounding on docs when working with agentic code editing (ex Roo, Claude Code).

Doctor uses crawl4ai to crawl websites, and then chunks and embeds them with langchain + litellm + openai, and finally stores all the vectors in duckdb. This allows your LLM to query the docs using semantic search over MCP, giving it grounded and up-to-date information for the things you're working on.

It requires an OpenAI key for the embedding process, but I'm working on giving users options in the future (different providers, local embedding using something like DPR or other transformer libs, etc.)

Show HN: Doctor – tool to crawl and index websites and MCP server for LLM agents

kixpanganiban — Tue, 13 May 2025 01:22:49 +0000

Article URL: https://github.com/sisig-ai/doctor

Comments URL: https://news.ycombinator.com/item?id=43968923

Points: 1

# Comments: 2

New comment by kixpanganiban in "Ask HN: Cursor or Windsurf?"

kixpanganiban — Mon, 12 May 2025 07:44:05 +0000

Interesting! Do you have an invite to spare? My email is in my bio

New comment by kixpanganiban in "Ask HN: SWEs how do you future-proof your career in light of LLMs?"

kixpanganiban — Tue, 17 Dec 2024 06:41:21 +0000

We had this same question when IDEs and autocomplete became a thing. We're still around today, just doing work that's a level harder :)

New comment by kixpanganiban in "Every V4 UUID"

kixpanganiban — Sat, 07 Dec 2024 00:48:13 +0000

Hehe 12345678-b00b-4123-b00b-b00b13551234

New comment by kixpanganiban in "/usr/bin/env -S uv run"

kixpanganiban — Thu, 21 Nov 2024 05:17:56 +0000

Curious - how many containers and machines images these days come with uv by default?

New comment by kixpanganiban in "Oasis: A Universe in a Transformer"

kixpanganiban — Sun, 03 Nov 2024 02:55:33 +0000

Right now it looks like Oasis is only trained on Minecraft. Imagine if it was trained in thousands of hours of other games as well, of different genres and styles.

Ostensibly, a game designer can then just "prompt" a new game concept they want to experiment with, and Oasis can dream it into a playable game.

For example, "an isometric top-down shooter, with Maniac mechanics, and Valheim graphics and worldcrafting, set in an ancient Nordic country"

And then the game studio will start building the actual game based on some final iteration of the concept prompt. A similar workflow already to concept art being "seeded" through Midjourney/SD/flux today.