Hacker News: s900mhz

New comment by s900mhz in "GLM-5V-Turbo: Toward a Native Foundation Model for Multimodal Agents"

s900mhz — Wed, 06 May 2026 02:36:42 +0000

Has 5.1 reliability improved? I would love to use it again. The inference was just too unreliable when it was first released.

New comment by s900mhz in "Anatomy of the .claude/ folder"

s900mhz — Fri, 27 Mar 2026 18:24:45 +0000

Yes this is the path I’m taking. Experiment, build your own toolbox whether it’s hand rolled skills or particular skills you pull out from other public repos. Then maintain your own set.

You do not want to log in one day to find your favorite workflow has changed via updates.

Then again this is all personal preference as well.

New comment by s900mhz in "Show HN: Lux – Drop-in Redis replacement in Rust. 5.6x faster, ~1MB Docker image"

s900mhz — Sun, 15 Mar 2026 22:41:53 +0000

Looks like the repo is very young.

First thing to do is try it out in a hobby project see how it works out!

New comment by s900mhz in "1M context is now generally available for Opus 4.6 and Sonnet 4.6"

s900mhz — Sat, 14 Mar 2026 03:31:20 +0000

I find it useful to make Claude track the debugging session with a markdown file. It’s like a persistent memory for a long session over many context windows.

Or make a subagent do the debugging and let the main agent orchestrate it over many subagent sessions.

New comment by s900mhz in "Claude Code On-the-Go"

s900mhz — Sun, 04 Jan 2026 22:59:02 +0000

I was doing the same, but recently I noticed that Claude now writes its plans to a markdown file somewhere nested in the ~/.claude/plans directory. It will carry a reference to it through compaction. Basically mimicking my own workflow!

This can be customized via a shell env variable that I cannot remember ATM.

The downside (upside?) is that the plan will not end up in your repo. Which sometimes I want. I love the native plan mode though.

New comment by s900mhz in "One year of keeping a tada list"

s900mhz — Sun, 28 Dec 2025 15:18:34 +0000

Why not? If one of your goals is to try to relax and/or meditate more, then I feel it’s a valid list entry.

It’s all a matter of perspective and personal goals, no?

New comment by s900mhz in "Claude in Chrome"

s900mhz — Sun, 21 Dec 2025 05:54:17 +0000

I personally replaced my playwright mcp with this. Seems to use less context and generally more reliable.

New comment by s900mhz in "Claude Advanced Tool Use"

s900mhz — Tue, 25 Nov 2025 05:08:46 +0000

This is a task I think is suited for a sub agent that is small in size. It can can take the context beating to query for relevant tools and return only what is necessary to the main agent thread.

New comment by s900mhz in "Study identifies weaknesses in how AI systems are evaluated"

s900mhz — Sat, 08 Nov 2025 18:57:33 +0000

When I have a bug I’m iterating on it’s much easier and faster to have it write out the playwright script. That way it does not have to waste time or tokens performing the same actions over and over again.

Think of it as TDD.

New comment by s900mhz in "Claude Skills"

s900mhz — Thu, 16 Oct 2025 18:11:32 +0000

Download the godot docs and tell the skill to use them. It won’t be able to fit the entire docs in the context but that’s not the point. Depending on the task it will search for what it needs

New comment by s900mhz in "Claude Sonnet 4.5"

s900mhz — Thu, 02 Oct 2025 00:45:21 +0000

I personally do not do any writes in parallel but parallel works great for read operations like investigating multiple failing tests.

New comment by s900mhz in "Unix philosophy and filesystem access makes Claude Code amazing"

s900mhz — Thu, 02 Oct 2025 00:42:48 +0000

Claude can then proceed to fix the issues for you

New comment by s900mhz in "Claude Sonnet 4.5"

s900mhz — Tue, 30 Sep 2025 04:42:32 +0000

You can use the built in task agent. When you have a plan and ready for Claude to implement, just say something along the line of “begin implementation, split each step into their own subagent, run them sequentially”

New comment by s900mhz in "Devious malware has jumped from Meta to Google Ads and YouTube to spread"

s900mhz — Fri, 26 Sep 2025 20:41:50 +0000

https://apps.apple.com/us/app/wipr-2/id1662217862

Wipr 2

It’s a one time paid app that works for both Mac and iOS

New comment by s900mhz in "AI overviews cause massive drop in search clicks"

s900mhz — Thu, 24 Jul 2025 19:28:15 +0000

Ahh thanks for the clarification, I misunderstood!

New comment by s900mhz in "AI overviews cause massive drop in search clicks"

s900mhz — Thu, 24 Jul 2025 00:35:15 +0000

It’s a feature of Kagi. Putting the question mark does invoke AI summaries.

https://help.kagi.com/kagi/ai/quick-answer.html

New comment by s900mhz in "What happens when people don't understand how AI works"

s900mhz — Mon, 09 Jun 2025 14:12:02 +0000

I personally love LLMs and use them daily for a variety of tasks. I really do not know how to “fix” the terminology. I agree with you that they are not thinking in the abstract like humans. I also do not know what else you would call “chain-of-thought”.

Perhaps “journaling-before-answering” lol. It’s basically talking out loud to itself. (Is that still being too anthropomorphic?)

Is this comment me “thinking out loud”? shrug

New comment by s900mhz in "Tracking Copilot vs. Codex vs. Cursor vs. Devin PR Performance"

s900mhz — Mon, 09 Jun 2025 14:03:03 +0000

I use Claude code daily at work, it writes all my PRs. It uses the GitHub cli to manage them.

Since all agents are able to use the terminal I suggest looking up the Gitlab CLI and have it use that. Should work locally and in runners.

New comment by s900mhz in "Cloudlflare builds OAuth with Claude and publishes all the prompts"

s900mhz — Wed, 04 Jun 2025 03:33:25 +0000

I’m not who you replied to but I keep functions small and testable paired with unit tests with a healthy mix of happy/sad path.

Afterwards I make sure the LLM passes all the tests before I spend my time to review the code.

I find this process keeps the iterations count low for review -> prompt -> review.

I personally love writing code with an LLM. I’m a sloppy typist but love programming. I find it’s a great burnout prevention.

For context: node.js development/React (a very LLM friendly stack.)

New comment by s900mhz in "xAI to pay telegram $300M to integrate Grok into the chat app"

s900mhz — Wed, 28 May 2025 19:11:31 +0000

Perplexity basically has this right? They let you buy things with a button in the chat itself!