Feed aggregator

Article URL: https://www.cnn.com/2026/03/09/economy/oil-price-shock

Comments URL: https://news.ycombinator.com/item?id=47309540

Points: 1

# Comments: 0

Article URL: https://www.worksinprogress.news/p/many-of-the-tastiest-vegetables-are

Comments URL: https://news.ycombinator.com/item?id=47309537

Points: 1

# Comments: 0

Article URL: https://github.com/cybrking/thr8

Comments URL: https://news.ycombinator.com/item?id=47309535

Points: 1

# Comments: 1

Article URL: https://hepa.pages.dev/

Comments URL: https://news.ycombinator.com/item?id=47309530

Points: 1

# Comments: 0

Article URL: https://www.rilldata.com/blog/building-an-agent-friendly-local-first-analytics-stack-with-motherduck-and-rill

Comments URL: https://news.ycombinator.com/item?id=47309523

Points: 1

# Comments: 0

I run a small data recovery business in Japan, and over the past year I've been building with AI coding tools like Claude Code, Cursor, and Cline.

One of my side projects is an overnight content pipeline for my business. It pulls RSS feeds, fetches source articles, generates posts with AI, scores them, and publishes them to WordPress without supervision.

The content is a bit niche: cybersecurity incidents for Japanese manufacturing companies in Aichi Prefecture — Toyota's home region — where older workflows like fax and password-protected ZIPs still haven't fully disappeared.

The physical setup is also a little ridiculous: a SwitchBot turns the PC on at 3am, Windows Task Scheduler starts the Python pipeline, and another scheduled task shuts the machine down when it's done.

Originally this was only meant to solve my own problem. But the more failure modes I found, the more features I kept adding.

What pushed me to build qzira was cost control.

The first lesson was operational: alerts don't help at 3am. What I needed wasn't another notification, but a kill switch outside the application — something that could stop requests before they reached the provider, regardless of what the agent decided to do.

The second lesson was more embarrassing: I had miscalculated Cloudflare KV write costs by 100x. Every request was triggering a KV put. Rewriting that path to batch via cron jobs reduced writes by about 99% and fixed the unit economics.

I also became much more conservative about model choice for production content after running a simple comparison on my own pipeline.

I ran 10 articles through Claude and didn't find any hallucinations.

Then I ran 1 article through gpt-4o-mini, and it immediately inverted the meaning of the source: it wrote "operations were suspended" where the original said "no impact on operations was confirmed."

To be fair, the pipeline was tuned around Claude, so I don't take this as a general statement about model quality. It may simply have been a prompt/model fit issue. But for me, it was enough to become much more conservative about where lower-cost models are allowed to touch production content.

Both problems pointed to the same conclusion: cost and policy enforcement belong at the infrastructure layer, not inside the application.

So I built qzira — a BYOK AI gateway in front of OpenAI, Anthropic, and Google AI. It adds gateway-level budget controls, hard stops, and routing by changing the base_url in tools like Claude Code or Cursor.

Stack: Cloudflare Workers, Hono, D1, KV, and Vectorize.

There's a free tier.

Happy to answer questions about the architecture, the cost mistake, the overnight pipeline, or the slightly absurd physical setup behind it.

Comments URL: https://news.ycombinator.com/item?id=47309509

Points: 1

# Comments: 0

Article URL: https://github.com/junianwoo/claude-toad

Comments URL: https://news.ycombinator.com/item?id=47309508

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=xe9OrBEn--w

Comments URL: https://news.ycombinator.com/item?id=47309489

Points: 1

# Comments: 0

Instead of giving LLM tools SSH access or installing them on a server, the following:

promptctl ssh user@server makes a set of locally defined prompts "appear" within the remote shell as executable command line programs.

For example:

# on remote host analyze-config --help Usage: analyze-config [OPTIONS] --path Prompt inputs: --all --path --opt --syntax --sec would render and execute the following prompt:

You are an expert sysadmin and security auditor analyzing the configuration file {{path}}, with contents: {{cat path}} Identify: {{#if (or all syntax) }}- Syntax Problems{{/if}} {{#if (or all sec) }}- Misconfigurations and security risks{{/if}} {{#if (or all opt) }}- Optimizations{{/if}} For each finding, state the setting, the impact, a fix, and a severity (Critical/Warning/Info). Nothing gets installed on the server, API keys never leave your computer, and you have full control over the context given to the LLM.

Github: https://github.com/tgalal/promptcmd/

Documentation: https://docs.promptcmd.sh/

Comments URL: https://news.ycombinator.com/item?id=47309480

Points: 1

# Comments: 0

Everyone you are selling to has access to Gemini, ChatGPT, or Claude.

If its a problem worth paying money to fix, they have already tried to solve their problem with one of these tools. They may already be using these tools (along with some automation tools) for a partial solution.

They have seen how it works, or doesn't work. They have some sense of the limitations of these tools.

Example: If you make a claim of "Run your blog on autopilot" you have to show me how you have solved the common issues that appear when trying to write a blog using AI. People have tried to create blogs with AI with ample, hand-written context and many writing samples. Don't tell me you've fixed this issue I could not fix - *show me* : An example, case study - anything like that. I want to read a real blog that your AI wrote (in this example.)

There is a tendency in AI marketing to gloss over the existing pain points/limitations of AI. I'm sure some of these common limitations are solvable - maybe with smart prompting or orchestration - but it's tiring to see so many landing pages/pitches hinge on "Just trust me bro" I'm sure you can wire up automations to make it all work, but if the quality isn't there with me manually using a chatbot, why would it be there with your tool?

I share this because this is what your potential customers are thinking.

Your product will always be judged against what the general chatbots can do. If you truly can get better results, prove it. If you are pitching automation or workflows around existing results, be clear about that (some people may be happy with that offering.)

Comments URL: https://news.ycombinator.com/item?id=47309479

Points: 1

# Comments: 0

Article URL: https://talonwatch.com

Comments URL: https://news.ycombinator.com/item?id=47309478

Points: 1

# Comments: 1

Article URL: https://www.notebookcheck.net/Apple-iPhone-Ultra-rumored-to-reach-2-900-price-tag.1245814.0.html

Comments URL: https://news.ycombinator.com/item?id=47309471

Points: 1

# Comments: 0

Article URL: https://pointersgonewild.com/2026-03-06-a-recursive-algorithm-to-render-signed-distance-fields/

Comments URL: https://news.ycombinator.com/item?id=47309461

Points: 1

# Comments: 0

Article URL: https://aarol.dev/posts/helloworld-print/

Comments URL: https://news.ycombinator.com/item?id=47309457

Points: 2

# Comments: 0

Article URL: https://kineticalpha.io/

Comments URL: https://news.ycombinator.com/item?id=47309450

Points: 1

# Comments: 1

Hi HN,

I built aeph, a small TUI app for jotting down notes and TODOs in Markdown without leaving the terminal.

The idea was inspired by Ephe (https://github.com/epheapp/ephe). I liked the concept of using Markdown as a kind of "paper notebook" for notes and tasks. It felt simple and flexible.

I wanted something similar but that could live entirely in the terminal, since I spend most of my time there.

aeph treats Markdown files like a notebook:

* write quick notes with md support * manage TODO checkboxes

It's written in Rust and uses a minimal TUI.

Interestingly, after a Japanese blog post introduced it, the npm package was downloaded a few thousand times. That made me curious how useful this idea might be for a broader developer audience.

I'd love to hear any thoughts or feedback.

Comments URL: https://news.ycombinator.com/item?id=47309447

Points: 1

# Comments: 0

Article URL: https://www.4rknova.com//blog/2025/06/22/mrs-fractal

Comments URL: https://news.ycombinator.com/item?id=47309445

Points: 1

# Comments: 0

Article URL: https://github.com/InferShield/infershield

Comments URL: https://news.ycombinator.com/item?id=47308737

Points: 1

# Comments: 0

Article URL: https://www.techradar.com/vehicle-tech/hybrid-electric-vehicles/get-a-grip-this-world-first-winter-tyre-can-automatically-deploy-studs-when-it-detects-freezing-road-conditions

Comments URL: https://news.ycombinator.com/item?id=47308713

Points: 1

# Comments: 0

Hi HN,

I’ve been building a small privacy tool called Exif Cleaner that removes metadata from images directly in the browser.

While working on it, I realized many people don’t actually know:

what EXIF metadata is

how much data photos contain

how location and device info can leak

So I started writing simple guides explaining EXIF metadata and how to remove it.

The goal is to make privacy concepts around photos easier to understand for non-technical users.

Current guides include topics like:

What EXIF metadata is

How to remove metadata from images

Why photos contain hidden information

Everything is free and the cleaner runs locally in the browser (no uploads).

Guide section: https://exif-cleaner.com/guide

Would love feedback on the guides or ideas for new ones.

Comments URL: https://news.ycombinator.com/item?id=47308708

Points: 1

# Comments: 0