Hacker News: sudb

New comment by sudb in "Artifacts: Versioned storage that speaks Git"

sudb — Thu, 16 Apr 2026 20:36:25 +0000

As cool as this technically is - who is the target market for this? I think people building coding agents and coding agent platforms are for the most part building on non-Cloudflare sandboxes, and can tolerate minutes of latency for setup.

I am not sure what people who roll their own in-house solutions for coding agents do, but I suspect that the easy path is still one of the many sandbox providers + GitHub.

I would love to find out who would use this & why!

New comment by sudb in "German Dog Commands"

sudb — Thu, 16 Apr 2026 20:02:15 +0000

I think this is a great idea in general - security through obfuscation, kinda.

New comment by sudb in "Stop Using Ollama"

sudb — Thu, 16 Apr 2026 06:53:18 +0000

What problem is it that you are confused isn't solved?

I think the codec analogy is neat but isn't the codec here llama.cpp, and the models are content files? Then the equivalent of VLC are things like LMStudio etc. which use llama.cpp to let you run models locally?

I'd guess one reason we haven't solved the "codec" layer is that there doesn't seem to be a standard that open model trainers have converged on yet?

New comment by sudb in "Stop Using Ollama"

sudb — Thu, 16 Apr 2026 06:42:28 +0000

seems pretty unrelated to the post?

also you might be the only person in the wild I've seen admit to this

New comment by sudb in "Phyphox – Physical Experiments Using a Smartphone"

sudb — Thu, 16 Apr 2026 01:04:28 +0000

Thanks for bringing this to light! Very interested to see if I can drive it!

New comment by sudb in "Phyphox – Physical Experiments Using a Smartphone"

sudb — Thu, 16 Apr 2026 01:02:19 +0000

If there was an app (web, mobile, desktop - whatever, open-source or not) that could take in smartphone-shot video of any 7-segment display and give you back some data of how the number on the displayed changed over time - is that not much, much easier than any of this? Does this also not generalize much better?

New comment by sudb in "Show HN: Continual Learning with .md"

sudb — Mon, 13 Apr 2026 21:53:22 +0000

I really like the simplicity of this! What's retrieval performance and speed like?

New comment by sudb in "I Just Want Simple S3"

sudb — Mon, 13 Apr 2026 21:50:17 +0000

I think the post author is mainly addressing self-hostable and/or open-source options here - otherwise I'd expect a whole host of other commercial storage providers to have been mentioned!

New comment by sudb in "I just want simple S3"

sudb — Mon, 13 Apr 2026 21:48:21 +0000

or cloudflare R2 for that matter (very useful for egress-heavy workloads for which it is ~free)

New comment by sudb in "Uncharted island soon to appear on nautical charts"

sudb — Sun, 12 Apr 2026 22:44:10 +0000

I assumed that satellite data would have been enough to know if something was an iceberg or land - wild that new land is still being discovered!

New comment by sudb in "Show HN: Claudraband – Claude Code for the Power User"

sudb — Sun, 12 Apr 2026 22:40:56 +0000

Are you asking if it is possible to be a power user of Claude Code? Because it very much is - what's the oxymoron?

New comment by sudb in "I run multiple $10K MRR companies on a $20/month tech stack"

sudb — Sun, 12 Apr 2026 21:04:21 +0000

Looks like this may be on the wrong parent post!

New comment by sudb in "Phyphox – Physical Experiments Using a Smartphone"

sudb — Sun, 12 Apr 2026 17:24:09 +0000

I think there's loads of scope to use phone cameras as dataloggers too, especially for older equipment that doesn't have an easy way to connect/export the data.

(I've been meaning for ages to write a piece of software that's able to extract change over time data from a video of a 7 segment display, like on a balance or a digital thermometer or something)

New comment by sudb in "Xkcd: Change in Slope"

sudb — Tue, 10 Mar 2026 00:56:50 +0000

there definitely are some areas in software-land where graphing data and/or directly eyeballing it genuinely helps to spot patterns where statistical methods might be cumbersome/tricky/otherwise annoying, like log analysis[1]

[1] https://jvns.ca/blog/2022/12/07/tips-for-analyzing-logs/

New comment by sudb in "Perhaps not Boring Technology after all"

sudb — Tue, 10 Mar 2026 00:48:44 +0000

This sharp uptick in LLM in-context "learning" capabilities means I'm more excited than ever to try to get to grips with "new" languages like Nim or Gleam (but worried that using LLMs to help me get to a working end state will rob me of some of the experience of learning).

New comment by sudb in "When does MCP make sense vs CLI?"

sudb — Mon, 02 Mar 2026 12:10:28 +0000

Every MCP vs CLI argument I've seen really glosses over _where_ the agent is running, and how that makes a difference. For individual users where you're running agents locally, I totally agree that CLIs cover the vast majority of use cases, where available.

I think something I've not seen anyone mention is that MCPs make much more sense to equip agents on 3rd party platforms with the tools they need - often installing specific CLIs isn't possible and there's the question of whether you trust the platform with your CLI authentication key.

How We Beat Claude Code on Terminal Bench

sudb — Thu, 17 Jul 2025 16:45:36 +0000

Article URL: https://www.enginelabs.ai/blog/engine-posts-sota-score-on-terminal-bench

Comments URL: https://news.ycombinator.com/item?id=44595285

Points: 3

# Comments: 0

Terminals for LLMs: A Halting Problem

sudb — Mon, 07 Jul 2025 18:59:43 +0000

Article URL: https://www.enginelabs.ai/blog/terminals-for-llms-a-halting-problem

Comments URL: https://news.ycombinator.com/item?id=44493571

Points: 2

# Comments: 0

New comment by sudb in "Show HN: Engine – A multi-LLM alternative to Codex"

sudb — Tue, 20 May 2025 13:56:11 +0000

We last submitted a SWE-Bench verified result in November 2024 - at the time I believe we were in the top 5 entrants.

We expect Engine to be as good as the other code-writing agents out there at the moment - we understand almost everyone in the space to be using very similar base models and agent scaffolding.

New comment by sudb in "Show HN: The Best Terminal Inspired Portfolio on the Internet™"

sudb — Tue, 20 May 2025 13:50:50 +0000

the closest I could get to getting your LLM to identify itself was as LaMDA, which makes me think this is probably a Gemma model - am I close?