Feed aggregator

We’re open-sourcing Browser Tools SDK: a small TypeScript package to give any AI agent a reliable way to control a real browser. With just a few lines of code, you can give any agent a production-ready browser harness

import { createAiSdkBrowserTools } from "libretto-browser-tools/ai-sdk"; import { LocalBrowserProvider } from "libretto-browser-tools"; const { tools } = createAiSdkBrowserTools(new LocalBrowserProvider()); const result = await generateText({ model: anthropic("claude-sonnet-4-5"), tools, prompt: "Go to Hacker News and tell me the top story", }); We built the Browser Tools SDK because access to a browser is one of the most important tools of a productive agent, but we found many of the existing tools lacking. Using the insights we gained building Libretto, we were able to build the SDK so that it’s significantly more context and cost-efficient than other tools:

We compared Browser Tools SDK against agent-browser, playwright-cli, and dev-browser on 26 live-site tasks with GPT 5.6 Sol (best results from 3 runs). Browser Tools tied for the best pass rate at 24/26 and came in about 55% cheaper per successful task than the next alternative ($0.106 vs $0.235) while also using far fewer tokens (1.45M vs 2.29M+). Benchmark methodology is in the repo.

The Browser Tools SDK only exposes 6 tools, of which only 2 are really important: `browser_snapshot` and `browser_exec`. The snapshot tool was designed from the ground up to be extremely context-efficient and provide agents the overview they need to know what to try next. The exec tool takes and executes raw Playwright code, of which the models have been trained on heaps of, and know how to use without much direction.

Works with AI SDK and Pi out of the box, plus a custom path for anything else. More frameworks are coming soon. You can also use any browser infra provider (Libretto Cloud, Kernel, Browserbase, etc.) or use `LocalBrowserProvider` to run Chromium locally (although that will have less anti-bot protections).

MIT licensed. Check out the docs: https://libretto.sh/docs/browser-tools/quickstart and join the Discord: https://discord.gg/NYrG56hVDt.

We’d love any and all feedback.

Comments URL: https://news.ycombinator.com/item?id=48998262

Points: 2

# Comments: 0

Article URL: https://zetadeck.com

Comments URL: https://news.ycombinator.com/item?id=48998261

Points: 1

# Comments: 0

Article URL: https://9to5mac.com/2026/07/21/ios-27-code-suggests-apple-could-restrict-leased-devices-after-missed-payments/

Comments URL: https://news.ycombinator.com/item?id=48998241

Points: 1

# Comments: 0

I've typically worked on longer and bigger software development engagements, but recently have picked up several small (but profitable) automation jobs via referral. My existing onboarding process is too much for these types of projects.

I’m curious what methodologies or tools (super curious about tooling) people use to help obtain the required context (asynchronously) to scope out a project successfully.

I feel like this is something that I’m good at and, probably, the most important input to delivering great outcomes for my clients. My level of engagement in this beginning phase is my biggest bottleneck, there is so much friction and energy required at the start. For reference on my existing pattern:

I tend to start with 1-1 conversations. This gives me a good starting point of context AND helps earn trust with stakeholders.

Then I move into 1 to many conversations which start to expose nuances as people talk to each other while I observe. This is also typically when I start to tease solutions and build up social excitement for the future state.

Both of these steps are very synchronous and time consuming. They also have valuable social benefits that remove friction when we move into implementation. But that doesn't seem as important in these more short term engagements. Curious what other folks are doing.

Comments URL: https://news.ycombinator.com/item?id=48998228

Points: 1

# Comments: 0

Article URL: https://twitter.com/KZouAPT/status/2079569508549673409

Comments URL: https://news.ycombinator.com/item?id=48998205

Points: 2

# Comments: 3

Easy to try even without an account! Would love all feedback.

Hey all, I'm building Meltbox as a better human in the loop system. It's a way to centralize all of the decisions you need to make into context-rich briefs in a system built for fast human review.

[Why]

Something major changed recently. Agents have gotten even better and even faster at building apps / sites / dashboards / etc... and that's changed my development workflow in a major way.

I often have my agents build an entire app/dashboard just to make ONE DECISION.

It's no longer just building the product. It's building throwaway internal apps on the fly to help me build the product.

So I built Meltbox as a home for those things. Some are throwaway briefs while others stick around as custom dashboards.

[How it Works]

- Go to https://meltbox.ai and an immediate workspace is spun up for you - Copy/paste the invite link to your agent (Claude Code / Codex / etc...) - The agent gets instructions on redeeming its access token and accessing your workspace - Ask your agent to push you a brief about any work you're doing - You review the brief in Meltbox while the agent stays connected live to it. - Make decisions, mark it up with comments, and write notes that all flow back to your agent. - Finally, Sign up to claim/save your workspace!

[Challenges]

In a world where it's easy to build anything, I question what it takes to build a SaaS tool worth buying than building yourself. Meltbox has required solving a lot of unique problems, not only in UX taste, but also: - Real time / web socket connections: agents live connected to the platform and your briefs - Secrets management: a secure architecture for managing API keys usable by apps and agents - Security / Permissions: RBAC permissions, content policies, domain safeguards and constant Fable / Sol audits - ...and so many more things.

These are not one-shottable. We're building the loops and burning the tokens to make Meltbox really good.

Comments URL: https://news.ycombinator.com/item?id=48998187

Points: 1

# Comments: 0

Article URL: https://moviesoon.eu/

Comments URL: https://news.ycombinator.com/item?id=48998179

Points: 3

# Comments: 0

Article URL: https://securityboulevard.com/2026/07/cisco-antares-a-new-family-of-open-source-inexpensive-compact-security-ai-models/

Comments URL: https://news.ycombinator.com/item?id=48998173

Points: 4

# Comments: 0

Academic IT and network services network Jisc resigns a collaborative MoU with partners from Australia, Canada, New Zealand and the US.

Here are hints and the answer for today's Wordle for July 22, No. 1,859.

Article URL: https://www.youtube.com/watch?v=61bgzr1LE8A

Comments URL: https://news.ycombinator.com/item?id=48996765

Points: 1

# Comments: 0

Article URL: https://thenextweb.com/news/tech-giants-hidden-off-balance-sheet-debt-ai

Comments URL: https://news.ycombinator.com/item?id=48996760

Points: 1

# Comments: 0

Article URL: https://www.newyorker.com/news/letter-from-the-uk/how-mudlarking-became-londons-hottest-hobby

Comments URL: https://news.ycombinator.com/item?id=48996747

Points: 1

# Comments: 0

Article URL: https://www.cnn.com/2026/07/21/world/france-becomes-the-first-eu-country-to-ban-social-media-for-children

Comments URL: https://news.ycombinator.com/item?id=48996706

Points: 2

# Comments: 0

Article URL: https://finance.yahoo.com/news/nearly-50-000-lake-tahoe-170925969.html

Comments URL: https://news.ycombinator.com/item?id=48996702

Points: 1

# Comments: 0

Article URL: https://www.nytimes.com/2026/07/21/technology/meta-ai-facebook-instagram-accounts.html

Comments URL: https://news.ycombinator.com/item?id=48996693

Points: 3

# Comments: 0

Article URL: https://linuxiac.com/firefox-best-privacy-extension-is-becoming-a-native-feature/

Comments URL: https://news.ycombinator.com/item?id=48996688

Points: 2

# Comments: 0

Article URL: https://www.quantamagazine.org/the-organisms-that-make-earths-harshest-places-home-20260720/

Comments URL: https://news.ycombinator.com/item?id=48996668

Points: 1

# Comments: 0

Article URL: https://apnews.com/article/ai-anthropic-copyright-settlement-claude-books-bartz-74b140444023898aeba8579b6e9f0d63

Comments URL: https://news.ycombinator.com/item?id=48996652

Points: 4

# Comments: 0