Reading up on Chrome

Chrome 151 introduces the <usermedia> HTML element to handle camera and microphone access declaratively, replacing script-triggered prompts and improving permission recovery rates.

A new tool called kage clones websites into offline, script-free static mirrors by rendering each page in headless Chrome and stripping JavaScript before saving.

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.

LLMs lower barriers to the web but inherit a stale, outdated model of the platform; new primitives like WebMCP and agent loops could reshape the web's future.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 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.

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.

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.

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.

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.

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.

The author expresses cautious optimism about small language models in browsers, highlighting privacy and low cost while noting concerns about calcification and standardization.

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.

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.

Chrome launches Skills, a feature to save, remix, and rerun AI prompts with one click across websites for personalized workflows.

Developer portals should optimize for AI coding agents by structuring documentation with llms.txt, token-efficient formatting, and capability signaling via skill.md.

The human.json protocol uses a JSON file and web of vouches between sites to let humans assert authorship and prove their content is not AI-generated.

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.

DenchClaw is an open-source AI CRM and workflow automation tool that runs locally on a Mac, using a Chrome profile to automate tasks like database queries, lead enrichment, and outreach.

Keep is a Chrome extension that saves URLs as clean Markdown for search and AI agent context via a linked account.

A security analysis using automated scanning found 36 malicious Chrome extensions that exfiltrate browsing data, affecting over 37 million users.

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.

LinkedIn scans browsers for over 2,900 Chrome extensions to detect and block scraping, automation, and data-harvesting tools used for spam.

The browser's sandbox, tested by decades of web abuse, provides a secure environment for AI agents using the File System Access API.

The author built FolioLM, a Chrome extension that collects browser sources and queries them with AI, transforming the browser into an information workshop.

Google is upgrading Gemini in Chrome with 'Skills' to let users define custom AI tasks, shifting from passive assistant to proactive browser agent.

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.

Vercel Labs' agent-browser is a Rust-based CLI tool enabling AI agents to automate browser interactions via commands and an MCP server.

Google Labs announces experimental 'Disco' browser on Chromium that uses Gemini 3 to generate interactive GenTabs like trip planners from chat prompts.

Agentic browsing tools are immature, splitting the web into human and AI lanes and forcing developers to redesign for both audiences.

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.

Manus AI launched a beta browser extension for Pro, Plus, and Team users, enabling automation in local Chrome/Edge browsers with authenticated sessions.