Hacker News

Article URL: https://pointc.co/know-your-point-c/

Comments URL: https://news.ycombinator.com/item?id=48308844

Points: 2

# Comments: 0

Article URL: https://mceglowski.substack.com/p/how-should-we-think-about-starship

Comments URL: https://news.ycombinator.com/item?id=48308832

Points: 1

# Comments: 0

Article URL: https://github.com/holepunchto/bare

Comments URL: https://news.ycombinator.com/item?id=48308825

Points: 1

# Comments: 0

Article URL: https://github.com/Diplomat-ai/diplomat-agent-ts

Comments URL: https://news.ycombinator.com/item?id=48308822

Points: 1

# Comments: 0

Article URL: https://www.theregister.com/cyber-crime/2026/05/27/supply-chain-brain-drain-npm-attacker-foolishly-leaks-own-github-private-token/5247424

Comments URL: https://news.ycombinator.com/item?id=48308801

Points: 1

# Comments: 0

Article URL: https://www.ssp.sh/brain/having-taste-with-ai/

Comments URL: https://news.ycombinator.com/item?id=48308798

Points: 2

# Comments: 0

Article URL: https://github.com/familiar-software/familiar

Comments URL: https://news.ycombinator.com/item?id=48308795

Points: 2

# Comments: 0

Article URL: https://www.construction-physics.com/p/where-are-the-economies-of-scale

Comments URL: https://news.ycombinator.com/item?id=48308784

Points: 3

# Comments: 0

Article URL: https://www.lrb.co.uk/the-paper/v24/n05/peter-campbell/why-does-it-take-so-long-to-mend-an-escalator

Comments URL: https://news.ycombinator.com/item?id=48308779

Points: 2

# Comments: 0

Article URL: https://mistral.ai/news/vibe-agent/

Comments URL: https://news.ycombinator.com/item?id=48308776

Points: 2

# Comments: 0

Article URL: https://ir.similarweb.com/news-events/press-releases/detail/139/manus-collaborates-with-similarweb-to-give-ai-agents-digital-marketing-insight

Comments URL: https://news.ycombinator.com/item?id=48308157

Points: 1

# Comments: 0

Article URL: https://rankops.net/nc/charlotte/

Comments URL: https://news.ycombinator.com/item?id=48308155

Points: 1

# Comments: 1

Hey, my name is Tyler, I made this. flashcardaudit.com is a tool that allows users to upload an Anki collection so that an AI auditor (Gemini 3.5 Flash) can review the factual correctness of each Anki card. I discovered that a non-trivial number of cards contain obvious factual errors; which means people are likely memorizing false information every day.

Here is a free example of what the audit results look like: https://flashcardaudit.com/results/sus8yj5wigtu

Video of me talking about it: https://www.youtube.com/watch?v=kW7vN_TBr60

You can learn more about my story here: https://flashcardaudit.com/about

More on how it works: https://flashcardaudit.com/how-it-works

FAQ: https://flashcardaudit.com/faq

Documentation and reading material here: https://flashcardaudit.com/library

Thanks for checking it out!

Comments URL: https://news.ycombinator.com/item?id=48308149

Points: 1

# Comments: 0

Hello everyone,

TL;DR Live demo: https://ag2b-example.vercel.app

Working on different projects, especially in B2B, I am getting the same request more and more often - "Add an AI feature, yesterday!" Most agent frameworks I found run the runtime on the server - and that pulls in a whole stack. Both backend and frontend have to learn how that runtime works and adopt its SDKs. The server turns into an orchestrator carrying most of the logic. For a feature that's really just "automate what the app already does, and add a brain," I kept asking the same thing: do we actually need all of that and how to ship these AI features faster?

AG2B (Agent to Browser) is my research on this question. The agent loop runs in the browser, where your app already is. Tools (the first primitive) are just your existing client functions (store actions, click handlers, whatever you already wrote), so there's no second tool registry and no contract to negotiate between the frontend and backend. The server shrinks (but not limited) to a thin proxy that keeps your API key out of the bundle plus optionally gates with user's permissions.

Because the loop runs where the UI lives, each turn can carry a fresh snapshot of what the user sees and can do right now. Scopes (the second primitive) re-inject live context on every iteration - into the system or user prompt, your choice. The model sees what's actually on screen right now. Nothing stale, zero roundtrips and clear context for the LLM.

On security take, does it secure? The LLM can only call the tools you delegate, and those tools are your existing handlers which already hit authenticated, permission-checked endpoints. A tool call goes through the same gates a user or frontend logic goes through.

What about existing server-side runtimes? Actually they can coexist. With AG2B you can write your own tool or custom provider which interacts with these endpoints.

There are some alternatives to run the agent in browser like working with DOM. That also works, but you pay for it a lot - parsing html and DOM operations are more expensive (tokens, bundle size) than telling LLM what it can do and provide a filtered context.

Two cases I built it for:

1. General app automation - a tiny LLM proxy + good prompting + your existing frontend logic wrapped with small primitives = the assistant can drive the app. 2. Domain tasks - e.g. "turn this question into a SQL query" over a big proprietary BI model. A fine-tuned/prompted model that knows the schema + an endpoint to hit it + one Tool - and you get the result.

It's plain TypeScript, provider-agnostic (OpenAI/Anthropic built in, or your own), with a hook system to observe/intercept every step (human-in-the-loop approvals, guardrails, RAG) and React bindings - headless hooks or a drop-in chat. Vue is coming.

There is also a plugin system. For example I created a WebMCP plugin which exposes your agent's tools through the browser API - which already could be tested in Chrome.

Repo: https://github.com/ag2b/ag2b Docs: https://ag2b.ai

Comments URL: https://news.ycombinator.com/item?id=48308148

Points: 1

# Comments: 0

Article URL: https://github.com/eudoxia0/hashcards

Comments URL: https://news.ycombinator.com/item?id=48308140

Points: 1

# Comments: 0

The AI craze, layoffs, and current job market should be a wake-up call for all of us. Companies HATE how much they pay us and how much leverage we’ve had all these years.

I believe the AI craze will slow and companies will realize they still need us. When we have leverage again, we should really form unions. Im not optimistic that we will but we should.

Comments URL: https://news.ycombinator.com/item?id=48308136

Points: 1

# Comments: 0

Article URL: https://www.minid.net/2019/8/12/in-defense-of-functional-css

Comments URL: https://news.ycombinator.com/item?id=48308123

Points: 1

# Comments: 0

Article URL: https://www.bbc.com/future/article/20260527-plots-love-letters-and-diplomacy-the-medieval-secrets-being-revealed-by-ai

Comments URL: https://news.ycombinator.com/item?id=48308122

Points: 1

# Comments: 0

Article URL: https://fergusfinn.com/blog/faster-than-speed-of-light/

Comments URL: https://news.ycombinator.com/item?id=48308116

Points: 1

# Comments: 0

Hi guys! I just developed this app for journaling. I tried to keep it very minimal, for people who like to stay focused on writing and not being distracted. It implements Ai prompts so you don't have to start with a blank page. The more you write the more diary will grow with you. The Ai will learn your themes and patterns and can make links in your story even after months. You will only receive one letter each day from the Ai so it's not invasive at all. Just you and your thoughts, at your own pace. I built it because I couldn't find anything that felt calm enough. Everything out there is trying to keep you engaged. This does the opposite. Give it a try and let me know what you think — I'd love some honest feedback from people who actually journal. nemejournal.app

Comments URL: https://news.ycombinator.com/item?id=48308115

Points: 1

# Comments: 0