---
title: 'In the News: November 10, 2026 (Extra 2)'
description: 'Browser vendors ship Modern Web Guidance 1.0, an agent skill pack whose own evals show large pass-rate gains on modern web checks.'
canonical_url: 'https://darkfactory.dev/news/2026-11-10-extra-2'
markdown_url: 'https://darkfactory.dev/news/2026-11-10-extra-2.md'
collection: news
date_published: '2026-11-10T16:10:00-05:00'
date_modified: '2026-11-10T16:10:00-05:00'
---

# In the News: November 10, 2026 (Extra 2)


A skill pack for coding agents, built by the Chrome team with Firefox, Edge and MDN, reached 1.0 today with its own before-and-after eval numbers. Separately, AWS made hardware-isolated sandboxes for agent workloads available on EKS, ECS and EC2 at no extra charge.

## 1. Browser vendors ship a skill pack for coding agents, with evals attached

**[Modern Web Guidance](https://github.com/GoogleChrome/modern-web-guidance)** · Google Chrome, with Microsoft Edge, Mozilla Firefox and MDN · GitHub, November 10, 2026

Modern Web Guidance is a set of agent skills and a local CLI. The agent searches about 200 hand-written guides on current HTML, CSS and browser APIs and loads the matching guide into its context. The 1.0 release, announced by Philip Walton on the [Chrome for Developers blog](https://developer.chrome.com/blog/modern-web-guidance-v1), adds a static legacy scan that flags old patterns to modernize, plus a public eval dashboard. The README states the problem directly: "Coding agents often default to older patterns because LLM training data contains vast amounts of legacy code."

The evals give agents realistic tasks on real web apps. Playwright and AST assertions grade the results, and each grader is calibrated to pass on a known-good solution and fail on the unmodified app. In the November 8 snapshot, on about 215 tasks, Claude Code on Opus 5.5 went from 57% to 88% of assertions passed with the guides. Codex CLI on GPT-6.1 Sol went from 54% to 82%, and Antigravity CLI on Gemini 4 Argon from 58% to 94%. These are the maintainers' own figures, and we have not rerun them.

Paul Irish of the Chrome team posted the release on [Hacker News](https://news.ycombinator.com/item?id=50443812) (144 points at about 16:00 EST). He wrote that the team prunes a guide when evals show models already know its material, and that the legacy scan covers about 20% of what the guides address. One commenter reported that the scan found only two issues in their code; Irish replied that it is tuned to avoid false positives. Telemetry is on by default and can be turned off with `DISABLE_TELEMETRY=1`.

**Why it matters:** This skill comes with an uplift measurement rather than only a description of what it should do, and the method is one a team can copy for its own skills: graders that must fail on the base app and pass on a reference solution, then guided and unguided runs compared. If agents write your frontend code, running the legacy scan on a repo you already know is a cheap first check of whether the guides find anything.

## 2. AWS puts each agent workload in its own Nitro VM, at no extra charge

**[Introducing AWS Nitro Sandboxes on Amazon EKS, ECS and EC2](https://aws.amazon.com/about-aws/whats-new/2026/11/nitro-sandboxes-eks-ecs-ec2/)** · Amazon Web Services · AWS What's New, November 9, 2026

AWS made Nitro Sandboxes available. They are lightweight VMs managed by the Nitro Hypervisor, and each EKS pod, ECS task or VM that uses one gets its own hardware isolation boundary on a shared EC2 instance. The announcement names agent code as a use: Nitro Sandboxes let you "safely execute LLM-generated code in an isolated environment of its own". AWS says enabling them in EKS or ECS is a one-line change, that secrets can be released only to sandboxes that pass cryptographic attestation, and that the feature runs on most Nitro-based instances in all commercial regions at no additional charge. Amazon Linux and Firecracker are supported natively. The page gives no performance figures beyond "near-native EC2 performance".

**Why it matters:** For a team already on EKS or ECS, a VM boundary per agent task becomes a setting rather than a separate system to run, if the one-line claim holds. The attestation step also gives a concrete way to keep credentials away from any sandbox that has not proven what it is running.