Reading up on browser-engineering
100 deep · digging since nov 25, 25
- Introducing the Safari MCP server for web developers
Apple released a Model Context Protocol server for Safari that lets AI agents inspect DOM, network requests, screenshots, and console output to debug websites autonomously.
- Introducing the <usermedia> HTML element | Blog
Chrome 151 introduces the <usermedia> HTML element to handle camera and microphone access declaratively, replacing script-triggered prompts and improving permission recovery rates.
- taken. — Since You Arrived Vol. IV
The page reveals the browser data websites silently collect—IP, timezone, GPU, fonts, battery level—without permission, highlighting fingerprinting prevalence.
- I Stored a Website in a Favicon
The author encoded a 208-byte HTML page into a 9x9 pixel favicon by storing RGB values and decoded it with JavaScript, demonstrating data storage in unintended places.
- Cloudflare teams up with Chrome, Firefox, and Edge on a privacy-first anti-bot protocol
Cloudflare, Mozilla, Google, and Microsoft are developing PACT, a privacy-first protocol to verify web traffic legitimacy without tracking users.
- Gemma 4 WebGPU Kernels - a Hugging Face Space by webml-community
Gemma 4 E2B runs locally in-browser via WebGPU, letting users prompt the model directly without server-side inference.
- Google Chrome update will fully close the door on ad blockers
Google Chrome is fully removing Manifest V2 support in upcoming releases, finally ending the last loophole that allowed ad blockers like uBlock Origin to work.
- The golden rule of Customizable Select
Safari 27's customizable select allows full visual styling of `<select>` elements without JavaScript, but requires text or accessible attributes on every option for usability, accessibility, and fallback compatibility.
- Google Chrome is killing all uBlock Origin bypasses, Microsoft Edge, Opera to follow - Neowin
Google Chrome removed the final feature flags that allowed uBlock Origin and other Manifest V2 extensions to function, with Edge and Opera expected to follow.
- Where my head is at with LLMs and the web
LLMs lower barriers to the web but inherit a stale, outdated model of the platform; new primitives like WebMCP and agent loops could reshape the web's future.
- Google tests sending Chrome users straight into AI Mode
Google tests a Chrome flag that redirects omnibox searches to AI Mode but insists it is just an exploration with no plans to change the default behavior.
- The future of the web is weirdly human - Jono Alderson
HTML-in-canvas enables rich 3D web experiences without sacrificing accessibility or machine readability, signaling a future where human-facing design is no longer limited by document formats.
- URL Source: https://www.reddit.com/r/SaaS/s/jfrZPkqgKO
Reddit blocks automated access from the scraping tool or script used to fetch this SaaS-related post, returning a network security block page.
- Build new features using built-in AI in Chrome | Blog
Chrome's built-in AI APIs (Summarizer, Prompt, Writer, Rewriter, Translator) let web apps run on-device models for tasks like headline generation, comment moderation, and translation, cutting cloud costs and keeping data private.
- Introduction
Partytown is a lazy-loaded library that relocates resource-intensive third-party scripts to a web worker to improve site performance by freeing the main thread.
- Announcing Web Serial Support in Firefox - Mozilla Hacks - the Web developer blog
Firefox 151 for Desktop adds Web Serial API support, enabling web apps to connect to serial hardware like microcontrollers and 3D printers.
- Don't Roll Your Own ... - Susam Pal
Websites should not override native browser behaviors for scrolling, navigation, text selection, or form controls, as doing so harms usability and breaks user expectations.
- Modern Web Guidance
Modern Web Guidance embeds web best practices and browser compatibility data into AI coding agents to improve web development workflows.
- What's new in web extensions: I/O 2026 recap | Blog
Google announced AI-driven developer growth, MWG skill, DevTools for agents, granular roles, enterprise publishing, and browser namespace support for Chrome extensions at I/O 2026.
- DOS Zone | Hacker News
Hacker News users discuss the DOS Zone web game archive, praising its nostalgia but reporting performance issues, crashes, and concerns about including non-DOS titles and copyright.
- browse.sh
browse.sh is a browser CLI that gives AI agents pre-built skills to automate websites, cutting token costs and enabling complex multi-site workflows.
- Install web apps with the new HTML install element | Blog
The new HTML `<install>` element lets developers add a browser-trusted install button for web apps without JavaScript, now in origin trial for Chrome and Edge.
- WebKit Features for Safari 26.5
Safari 26.5 introduces the :open pseudo-class, element-scoped random(), color-interpolation for SVG gradients, ToggleEvent.source, and the Origin API, along with 63 bug fixes.
- Arthur Cornil
Many SaaS apps don't need servers; shipping the full backend to the browser via WASM, SQLite, and OPFS enables local-first single-user apps.
- Multi-stroke text effect in CSS
Stacking CSS text-stroke layers with varying widths creates a multi-stroke text effect, though browser rendering differences (Firefox smoother than Chrome/Safari) and poor performance make it unsuitable for production.
- How Anthropic’s Mythos has rewritten Firefox’s approach to cybersecurity
Anthropic's Mythos model unearthed hundreds of high-severity Firefox bugs, including sandbox vulnerabilities that human researchers rarely found.
- Behind the Scenes Hardening Firefox with Claude Mythos Preview - Mozilla Hacks - the Web developer blog
Mozilla details how agentic AI harnesses using Claude Mythos Preview found 271 real Firefox security bugs by dynamically creating reproducible test cases and dismissing false positives.
- Google’s Prompt API
Google's Prompt API ships as a Chrome-only web standard requiring users to accept Google's use policy and download Gemini Nano without permission, drawing opposition from Mozilla and WebKit.
- Tool: Redis Array Playground
An interactive playground lets users try experimental Redis array commands via WASM, including ARGREP for server-side regex matching.
- The duality of language models in the browser - daverupert.com
The author expresses cautious optimism about small language models in browsers, highlighting privacy and low cost while noting concerns about calcification and standardization.
- Mozilla pushes back against Google's Prompt API
Mozilla opposes Google's Prompt API in Chrome, arguing it threatens web interoperability and neutrality by tying the API to Google's Gemini Nano model and policies.
- The Bloomberg Terminal Is Getting an AI Makeover, Like It or Not
Bloomberg is testing ASKB, an AI chatbot for its Terminal, to help traders synthesize vast data via natural language, though it may challenge traditionalists.
- Installing every* Firefox extension
A developer scraped all 84,000+ Firefox extensions from AMO, installed them, and documented the resulting browser performance and technical challenges.
- Google upgrades AI Mode in the Chrome browser
Google updates AI Mode in Chrome to open web pages side-by-side with search, enabling contextual follow-up questions and cross-tab search across tabs, images, and files.
- Browser Run: give your agents a browser
Cloudflare renames Browser Rendering to Browser Run, adding Live View, Human in the Loop, CDP access, WebMCP support, session recordings, and 4x higher concurrency for AI agents.
- Name-only @container queries: A solution to the naming wars
Safari 26.4 ships name-only @container queries, resolving naming conflicts by allowing queries without explicit container names.
- Turn your best AI prompts into one-click tools in Chrome
Chrome launches Skills, a feature to save, remix, and rerun AI prompts with one click across websites for personalized workflows.
- Making WebAssembly a first-class language on the Web
A proposal to make WebAssembly a first-class language on the web using the Component Model for direct Web API access, eliminating JavaScript glue code.
- Chrome DevTools MCP (2025)
Chrome DevTools MCP now allows coding agents to connect to active browser sessions, enabling direct access to authenticated sessions and active debugging for AI-assisted workflows.
- Hardening Firefox with Anthropic's Red Team
Anthropic used Claude LLM to find 22 Firefox vulnerabilities, but exploits only worked in a test environment without sandbox, highlighting defense in depth.
- Meet WebKit for SwiftUI - WWDC25 - Videos - Apple Developer
Apple introduces a SwiftUI-native WebKit API to simplify embedding and controlling web content across Apple platforms.
- Claude for Chrome Extension Internals (v1.0.56)
The Claude for Chrome extension uses a Manifest V3 architecture with a React side panel, an agentic tool-use loop, and extensive Chrome DevTools Protocol integration for AI-driven browser automation.
- Show HN: X86CSS – An x86 CPU emulator written in CSS
A developer created a fully functional x86 CPU emulator using only CSS, demonstrating Turing completeness of CSS, though it currently only runs in Chromium-based browsers.
- BU Agent API (Experimental) - Browser Use Cloud
Browser Use Cloud's new experimental BU Agent API enables automated browser tasks such as data extraction, form filling, and multi-step workflows through a Python SDK.
- Perplexity tests Messages integration and usage credits
Perplexity is developing a Mac Messages connector for its Comet browser and testing a usage-credits system for Pro users to buy extra capacity without upgrading.
- Dictionary Compression is finally here, and it's ridiculously good
Dictionary compression using Zstandard or Brotli can reduce web traffic by up to 90% for returning users by leveraging previously downloaded content as a shared dictionary.
- Zero-day CSS: CVE-2026-2441 exists in the wild
Google patched a high-severity use-after-free vulnerability in Chromium's CSS engine exploited in the wild as a zero-day.
- keep.md - Chrome Web Store
Keep is a Chrome extension that saves URLs as clean Markdown for search and AI agent context via a linked account.
- Chrome extensions spying on users' browsing data
A security analysis using automated scanning found 36 malicious Chrome extensions that exfiltrate browsing data, affecting over 37 million users.
- Perplexity Comet: A Reversing Story
The article provides a technical deep-dive into Perplexity's agentic browser Comet, detailing its dual-channel communication, hard boundaries, and page representation for AI interaction.
- GitHub - needle-tools/md-browse: Markdown Browser – See the web like an AI does
md-browse is a markdown-first browser built with Electrobun that prioritizes markdown content and converts HTML to clean markdown using Turndown.
- LinkedIn checks for 2953 browser extensions
LinkedIn scans browsers for over 2,900 Chrome extensions to detect and block scraping, automation, and data-harvesting tools used for spam.
- One Human + One Agent = One Browser From Scratch
A single Codex CLI agent built a basic HTML+CSS browser in three days with 20,000 lines of Rust, challenging the need for large multi-agent projects.
- the browser is the sandbox
The browser's 30-year-old sandbox model can be adapted to safely run AI agents locally, though gaps remain in network isolation and malicious file creation.
- The browser is the sandbox
The browser's sandbox, tested by decades of web abuse, provides a secure environment for AI agents using the File System Access API.
- If NotebookLM was a web browser
The author built FolioLM, a Chrome extension that collects browser sources and queries them with AI, transforming the browser into an information workshop.
- The Overcomplexity of the Shadcn Radio Button
The Hacker News discussion argues that the Shadcn radio button's complexity reflects deeper issues with web UI primitives, despite some defending its necessity for customization.
- jQuery 4 | Hacker News
jQuery 4.0.0 final release is the first major version in nearly a decade, dropping IE<11 support and adding Trusted Types and CSP support.
- Consent-O-Matic
Consent-O-Matic is a browser extension that automatically handles cookie consent banners based on user preferences, saving time and reducing repetitive clicks.
- Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)
ChartGPU is a WebGPU-powered charting library that renders 1 million data points at 60 frames per second in the browser.
- Chrome’s Gemini is getting “Skills” as it moves toward becoming a full AI agent
Google is upgrading Gemini in Chrome with 'Skills' to let users define custom AI tasks, shifting from passive assistant to proactive browser agent.
- Show HN: Webctl – Browser automation for agents based on CLI instead of MCP
Webctl offers a CLI-based browser automation tool for AI agents that outperforms MCP-based tools with lower cost and higher quality.
- Just the Browser
Just the Browser provides configuration files and scripts to disable AI features, telemetry, sponsored content, and other annoyances in Chrome, Edge, and Firefox using group policies.
- Google tests Gemini Auto Browse tool for Chrome users
Google is testing an Auto Browse tool for Gemini that lets the AI autonomously manage tabs and browsing in Chrome, likely as a premium Ultra feature.
- Text-based web browsers
Text-based browsers like Lynx and w3m are increasingly incompatible with modern web technologies, leading users to advocate for alternatives like Gemini.
- Agent - 🤘 Stagehand
Stagehand's `agent()` function enables fully autonomous browser workflows using AI, with support for multiple agent modes, custom tools, streaming, and structured output.
- GitHub - vercel-labs/agent-browser: Browser automation CLI for AI agents
Vercel Labs' agent-browser is a Rust-based CLI tool enabling AI agents to automate browser interactions via commands and an MCP server.
- This game is a single 13 KiB file that runs on Windows, Linux and in the Browser
A developer created a 13 KiB snake game that runs natively on Windows, Linux, and in the browser as a single polyglot file by implementing the game three times.
- Show HN: An interactive guide to how browsers work
An interactive guide explains browser mechanics from URL parsing to TCP connections and HTTP requests, aiming to build intuitive understanding.
- A first look at the Web Install API – Frontend Masters Blog
The Web Install API enters origin trial in Chromium, allowing any website to install a progressive web app with a single button click.
- How I browse the web in 2026
A developer details their 2026 web browsing setup using Firefox on separated work and personal devices, with curated extensions, self-hosted RSS, and bookmarks for focused reading.
- Use Claude Code with Chrome (beta) - Claude Code Docs
Claude Code integrates with the Chrome extension to enable browser automation, live debugging, and multi-step workflows from the CLI or VS Code.
- The HTML Elements Time Forgot - HTMHell
The article catalogs long-obsolete HTML elements like `<xmp>`, `<spacer>`, `<keygen>`, and `<isindex>`, explaining their original purposes and why they were abandoned.
- URL Pattern API - Web APIs
The URL Pattern API provides a standardized syntax and JavaScript interface for matching URLs using patterns with named groups, wildcards, and regex, based on path-to-regexp.
- Noclip.website – A digital museum of video game levels
Noclip.website lets users explore rendered 3D video game levels from dozens of classic titles directly in a browser using WebGL and WebGPU.
- Is Mozilla trying hard to kill itself?
Mozilla CEO Enzor-DeMeo stated blocking adblockers in Firefox could bring in $150 million but said it feels off-mission, sparking backlash from users.
- Mozilla appoints new CEO Anthony Enzor-Demeo
HN commenters largely criticize Mozilla's new CEO, his MBA background, and the company's pivot to AI, arguing it undermines Firefox's only remaining differentiator of trust.
- Just a moment...
Cloudflare's 2025 Year in Review report reveals continued internet traffic growth, generative AI bot surge, and increased adoption of post-quantum encryption and IPv6 across its global network.
- Google announces experimental ‘Disco’ browser with ‘GenTabs’
Google Labs announces experimental 'Disco' browser on Chromium that uses Gemini 3 to generate interactive GenTabs like trip planners from chat prompts.
- Should CSS be Constraints?
CSS layout complexity makes both rule-based and constraint-based systems unpredictable; newer modes like Flexbox and Grid improve the situation better than replacing CSS entirely.
- What Is Agentic Browsing and Why Are AI Browsers Not Replacing Chrome Yet? - Bloomberg
Agentic browsing tools are immature, splitting the web into human and AI lanes and forcing developers to redesign for both audiences.
- Google Online Security Blog: Architecting Security for Agentic Capabilities in Chrome
Chrome introduces layered defenses like the User Alignment Critic and origin-based gating to protect users from indirect prompt injection during agentic browsing with Gemini.
Takes
http://knockoff.co is at around 75,000 daily active users! 🤯 a *browser extension*. absolutely bonkers.
@Shpigford
🖨️ It took over a year and lots of help from Claude Code but I have now been able to create a real dot matrix printer on the web that prints from Windows 3.11 in your browser: Here's the test page:
@levelsio
The things that Codex can do with browser use blow my mind
@petergyang
i built localterm run a terminal in your browser npx localterm@latest start
@aidenybai
@aviflombaum @Cloudflare today's victims were headless browser APIs. switched to cloudflare's browser run.
@Shpigford
Brave just added a feature people have wanted for YEARS. Containers. And most users don’t realize how powerful this is. Here’s why it changes how you browse 👇
@Techjunkie_Aman
When Brave users told us they'd pay for a minimalist version of our browser, we listened. Brave Origin, now in Nightly, is a paid version of our browser for users who don't need all of Brave’s features but still want its leading privacy and ad blocker: https://account.brave.com/?intent=checkout&product=origin
@BraveNightly
PSA: You can vibe code your own "New tab" page in Chrome. I have turned mine into the ultimate solution to the "too many tabs" problem - See all your tabs with clear titles, grouped by domain - Closing any tab gives you "swoosh" sound and confetti effect 🎊 - "Easy wins" grouped together: homepages, localhost tabs... batch-close them with one click - Duplicate tabs detected; close duplicates with one click - For tabs you're not done with, save it for later in a checklist This is the Marie Kondo method for browser tabs Open-sourced the code below
@zarazhangrui
So @brave just launched Brave Origin, a bare-bones, privacy-first browser without any ads, AI, or web3. It's $60. Unless you're on Linux, in which case it's free. Pic related (arch btw)
@o7laurence
Introducing /cookie-sync Run browser tasks in the cloud with all your authenticated accounts, powered by @browserbase Watch as my agent: - uploads my local Chrome cookies - injects them into a remote browser - goes to Forkable & chooses my Friday lunch for me
@shreypandya
OMG you guys, this is incredible! This is using Google's new WebMCP function to control your browser, but not only is it lightning fast, but its unique because it is using your main Chrome instance. Not some sandboxxed Playwright instance that doesn't want to remember your sessions, cookies, or passwords. Your real Chrome instance. It's incredible. You need to enable: chrome://inspect/#remote-debugging Also, it doesn't even require a skill to use. It just works. I'm thinking about making one anyway. I'm telling you, download this and try it. This is my new daily for sure.
@LLMJunky
Introducing the new dev-browser cli. The fastest way for an agent to use a browser is to let it write code. Just `npm i -g dev-browser` and tell your agent to "use dev-browser"
@sawyerhood
more importantly, Chrome 146 finally has vertical tabs! Enable it in chrome://flags, relaunch, then right-click on the tab bar and set it to vertical
@Stammy
Chrome 146 includes an early preview of WebMCP, accessible via a flag, that lets AI agents query and execute services without browsing the web app like a user.Services can be declared through an imperative navigator.modelContext API or declaratively through a form. pic.twitter.com/UaUplZ8Q28
@firt
𝚊𝚐𝚎𝚗𝚝-𝚋𝚛𝚘𝚠𝚜𝚎𝚛 v0.9.1--𝚊𝚕𝚕𝚘𝚠-𝚏𝚒𝚕𝚎-𝚊𝚌𝚌𝚎𝚜𝚜open local PDFs and HTML files via 𝚏𝚒𝚕𝚎:// URLs-𝙲 / --𝚌𝚞𝚛𝚜𝚘𝚛capture clickable divs with 𝚘𝚗𝚌𝚕𝚒𝚌𝚔 or 𝚌𝚞𝚛𝚜𝚘𝚛:𝚙𝚘𝚒𝚗𝚝𝚎𝚛 in snapshots𝚗𝚙𝚖 𝚒 -𝚐 𝚊𝚐𝚎𝚗𝚝-𝚋𝚛𝚘𝚠𝚜𝚎𝚛
@ctatedev
okay, so @vercel's new agent browser is pretty stinkin great.https://t.co/FfyoARDd86substantially faster than playwright and is clearly optimized for AI use.well done, folks!
@Shpigford
Claude in Chrome runs in a side panel that stays open as you browse, working with your existing logins and bookmarks.Try it out: https://t.co/izQBKCBuWd
@claudeai
Here's why: https://t.co/ADtxy9msf4
@MishaalRahman