Hacker News Reader

SparkyMcUnicornon I built an AI receptionist for a mechanic shop

I don't think it's a bug, it's an extra hoop to jump through.

SparkyMcUnicornon Show HN: Tmux-IDE, OSS agent-first terminal IDE

It sounds like maybe you haven't seen agent teams mode, which this project is using.

https://code.claude.com/docs/en/agent-teams

SparkyMcUnicornon Mouser: An open source alternative to Logi-Plus mouse software

I use the g305, and have to replace it roughly once a year because the scroll button eventually stops working. I've been through 5-6 of them. Regularly on sale for around $25.

Wish I could find a better mouse I like as much as this one that I didn't need to pay a yearly subscription for. It's just the right size, lightweight, wireless, and being able to store the customizations to on board memory is nice.

SparkyMcUnicornon Mullvad VPN: Banned TV Ad in the Streets of London [video]

> By providing an example of similar articles about other locations I was showing that this alone is not particularly strong evidence.

How does linking similar articles help demonstrate that? That doesn't seem like very strong (counter?)evidence.

SparkyMcUnicornon Mullvad VPN: Banned TV Ad in the Streets of London [video]

This is whataboutism. It doesn't address any of the points made above.

SparkyMcUnicornon Halt and Catch Fire: TV’s best drama you’ve probably never heard of (2021)

Somewhat mysteriously, the Linux ISOs for this show have seen a sudden spike in activity.

SparkyMcUnicornon An Update on Heroku

> lead to customers moving elsewhere

Since they're no longer accepting new enterprise clients, maybe this is intentional.

SparkyMcUnicornon Netbird – Open Source Zero Trust Networking

I've done this a few different ways.

I started with a docker container that connected to both the VPN provider and tailscale. Now OPNSense is handing a few connections to the VPN provider at a couple locations around the world, and enforcing external traffic to be routed to the VPN connections via VLAN tags (untagged has direct internet access).

Using the VPN provider can either be adding a VLAN tag to a machine/container or connecting to a "vpn-{location}" tailscale exit node.

SparkyMcUnicornon Show HN: macOS menu bar app to track Claude usage in real time

Have you seen aichat[0]?

Doesn't support the "from anywhere" part, but the resume strategies are pretty cool.

[0] https://github.com/pchalasani/claude-code-tools#aichat-sessi...

SparkyMcUnicornon Show HN: Local Privacy Firewall-blocks PII and secrets before ChatGPT sees them

I configure permission settings within projects.

https://code.claude.com/docs/en/settings#permission-settings

SparkyMcUnicornon GPT-5.2

If the pretraining rumors are true, they're probably using continued pretraining on the older weights. Right?

SparkyMcUnicornon Cloudflare Sandbox SDK

To me, the docs answer it pretty clearly. The defined directories persist until you destroy().

The part that's unclear to me is how billing works for a sandbox's disk that's asleep, because container disks are ephemeral and don't survive sleep[2] but the sandbox pricing points you to containers which says "Charges stop after the container instance goes to sleep".

https://developers.cloudflare.com/sandbox/concepts/sandboxes...

[2] https://developers.cloudflare.com/containers/faq/#is-disk-pe...

SparkyMcUnicornon Claude Code 2.0

If you turn off "Help improve Claude" you will never get this prompt (I never do).

https://claude.ai/settings/data-privacy-controls

SparkyMcUnicornon Context is the bottleneck for coding agents now

> For finicky issues like that I often find that, in the time it takes to create a prompt with the necessary context, I was able to just make the one line tweak myself.

I don't run into this problem. Maybe the type of code we're working on is just very different. In my experience, if a one-line tweak is the answer and I'm spending a lot of time tweaking a prompt, then I might be holding the tool wrong.

Agree on those terms being relative. Maybe a better way of putting it is that I'm very comfortable putting my name on it, deploying to production, and taking responsibility for any bugs.

SparkyMcUnicornon Context is the bottleneck for coding agents now

I'll stop you right there. I've been using Claude Code for almost a year on production software with pretty large codebases. Both multi-repo and monorepo.

Claude is able to create entire PRs for me that are clean, well written, and maintainable.

Can it fail spectacularly? Yes, and it does sometimes. Can it be given good instructions and produce results that feel like magic? Also yes.

SparkyMcUnicornon Show HN: The text disappears when you screenshot it

Yes: https://jsfiddle.net/kx6stbcL/

SparkyMcUnicornon Microsoft Favors Anthropic over OpenAI for Visual Studio Code

I decided to give perplexity another try a few days ago, and it still seems to hallucinate things. Given the same exact tasks/prompts both Claude and Chatgpt got the facts correct.

SparkyMcUnicornon macOS Tahoe

You can get a 128GB MacBook for around $4300. With a 4TB SSD it'll be around $5k I think.

https://www.apple.com/shop/product/G1FW7LL/A/Refurbished-16-...

SparkyMcUnicornon Qwen3-Next

This would be a valuable benchmark if it included languages other than Java, and let me see which models are best at the languages I work with.

My real-world usage does not line up with these results, but I'm not working with Java.

SparkyMcUnicornon QGIS is a free, open-source, cross platform geographical information system

A couple months ago I needed some info about a home/property. The county publishes a zip file of "geo data", and after downloading it I asked Claude Code how I can work with the files. It recommended QGIS, but I was lost when I opened it and searched GitHub to find this MCP server[0].

20 minutes later it had every piece of data I could possibly want placed onto the parcel on the map. It felt magical.

[0] https://github.com/jjsantos01/qgis_mcp/

SparkyMcUnicornon Claude now has access to a server-side container environment

I'd believe a lot of other claims before believing model degradation was happening.

- They admittedly go off of "vibes" for system prompt updates[0]

- I've seen my coworkers making a lot of bad config and CLAUDE.md updates, MCP server span, etc. and claiming the model got worse. After running it with a clean slate, they redacted their claims.

[0] https://youtu.be/iF9iV4xponk?t=459

SparkyMcUnicornon Claude now has access to a server-side container environment

"or other factors" is pretty catch-all in my opinion.

> I don't know how they could have phrased it that closed the door on this concern

Agreed. A full legal document would probably be the only way to convince everyone.

SparkyMcUnicornon Show HN: Attempt – A CLI for retrying fallible commands

`dotry`

SparkyMcUnicornon Experimenting with Local LLMs on macOS

This should be what you're looking for. It doesn't utilize the GPU, but WebGL support is in the TODOs.

https://github.com/ngxson/wllama

https://huggingface.co/spaces/ngxson/wllama

SparkyMcUnicornon Experimenting with Local LLMs on macOS

Yes. MLC's inference engine runs on WebGPU/WASM.

https://github.com/mlc-ai/web-llm-chat

https://github.com/mlc-ai/mlc-llm

https://github.com/mlc-ai/web-llm

SparkyMcUnicornon GLM 4.5 with Claude Code

When it comes to "real-world development scenarios" they claim to be closer to Sonnet 4.

This is the data for that claim: https://huggingface.co/datasets/zai-org/CC-Bench-trajectorie...

SparkyMcUnicornon MentraOS – open-source Smart glasses OS

Very interesting. Thanks for the correction/info!

SparkyMcUnicornon MentraOS – open-source Smart glasses OS

AOSP (or even a minimal fork) is way too heavy to be running on the glasses. It looks like the firmware is quite minimal and the "OS" is the app.

https://github.com/Mentra-Community/MentraOS/tree/main/mcu_c...

SparkyMcUnicornon MentraOS – open-source Smart glasses OS

The "Even Realities G1" and "Vuzix Z100" options listed in the compatibility doc look interesting.

https://github.com/Mentra-Community/MentraOS/blob/main/glass...

https://www.evenrealities.com/g1

https://www.vuzix.com/products/z100-smart-glasses

SparkyMcUnicornon MentraOS – open-source Smart glasses OS

It looks like it really is open source. Even the "cloud" components appear to be open source, and are in this repo.

https://docs.mentra.glass/contributing