---
type: "post"
id: "google-chrome-s-modern-web-guidance-packages-expert-727856"
slug: "google-chrome-s-modern-web-guidance-packages-expert-727856"
title: "Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that"
description: "Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that coding agents can retrieve through a CLI."
retrieval_nugget: "Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that coding agents can retrieve through a CLI. It gives teams a practical way to steer generated web code toward current browser capabilities, accessibility, and performance patterns rather than relying on model memory alone."
published_at: "2026-11-11"
updated_at: "2026-11-11"
record_date: "2026-11-11"
date_kind: "published_at"
topics: ["google","coding-agents"]
entities: ["Google Chrome","Modern Web Guidance","Microsoft Edge","Mozilla Firefox","MDN"]
related_patterns: ["design-systems-become-agent-context"]
novelty_label: "new_fact"
what_changed: null
prior_refs: []
source_url: "https://github.com/GoogleChrome/modern-web-guidance-src"
source_urls: ["https://github.com/GoogleChrome/modern-web-guidance-src"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:google-chrome-s-modern-web-guidance-packages-expert-727856"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/google-chrome-s-modern-web-guidance-packages-expert-727856/","markdown":"https://newruntime.com/posts/google-chrome-s-modern-web-guidance-packages-expert-727856.md","json":"https://newruntime.com/posts/google-chrome-s-modern-web-guidance-packages-expert-727856.json"}
markdown_url: "https://newruntime.com/posts/google-chrome-s-modern-web-guidance-packages-expert-727856.md"
---

# Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that

## Retrieval answer

Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that coding agents can retrieve through a CLI. It gives teams a practical way to steer generated web code toward current browser capabilities, accessibility, and performance patterns rather than relying on model memory alone.

## Summary
Google Chrome’s Modern Web Guidance packages expert-curated web-platform recommendations as skills that coding agents can retrieve through a CLI. It gives teams a practical way to steer generated web code toward current browser capabilities, accessibility, and performance patterns rather than relying on model memory alone.

## What happened
The GoogleChrome/modern-web-guidance-src repository describes itself as the source repository for Modern Web Guidance, a set of skills that puts web-platform expertise, best practices, and browser-compatibility data into coding agents’ context. The stated goal is to steer agents away from legacy patterns and toward modern web-platform solutions. The project is stewarded by the Google Chrome, Microsoft Edge, Mozilla Firefox, and MDN teams alongside the web-development community.

The repository gives an installation command: `npx modern-web-guidance@latest install`. It runs an interactive wizard. It also describes ways to try the CLI without installing the package: `search` finds relevant guides, `retrieve` returns a guide by ID, and `legacy-scan` scans a codebase for legacy patterns. The repository’s examples search for guidance on animating a dialog, retrieve the `animate-to-from-top-layer` guide, and scan the current directory with `npx modern-web-guidance@latest legacy-scan .`.

The guidance spans multiple disciplines. The listed areas include user experience, CSS layout, performance, forms and UI, accessibility, and built-in AI. Examples include view transitions, container queries, `oklch`, `scheduler.yield`, the Popover API, dialogs, keyboard operability, and local browser APIs for translation, summarization, and language detection. The README describes the guides as token-efficient and says evaluations are used to prune content that models already know. It identifies a full list covering 250 modern web features, including 140 CSS and layout features.

Google Chrome’s repository argues that models can produce bloated JavaScript where native browser capabilities can do the job. It also says a model may know an API exists without having dense, real-world implementation patterns needed for production-ready code. The CLI is intended to return targeted guidelines into an agent’s context window, rather than requiring a developer to paste broad reference material into every request. The project links to documentation, an evaluation dashboard, the source code, and an issue tracker.

This is not a browser feature that changes how a deployed application runs by itself. It is a maintained layer of instructions and retrieval tools around an agent’s implementation process. In practice, a developer chooses a task, retrieves focused guidance, and still decides whether the resulting code belongs in the product. That distinction matters: Modern Web Guidance aims to improve the context available to a coding agent, not replace browser testing, code review, or compatibility decisions.

## What the archive adds
New Runtime’s earlier coverage on May 20 described Modern Web Guidance as expert-vetted skills intended to help AI coding agents use modern web APIs rather than patterns drawn from older training data. This is a new fact on a known story: the repository documents the CLI workflow, evaluation framing, project stewardship, and stated coverage of 250 features.

## Trend signal
This supports the pattern of [design systems becoming executable agent context](https://newruntime.com/patterns/design-systems-become-agent-context/): expertise is delivered as retrievable, task-relevant guidance inside an agent workflow instead of being left in static documentation. The material is broader than a design system, but the mechanism is similar: maintained knowledge becomes operational context at the point of work.

## Why now
The repository frames the problem as a mismatch between the large volume of legacy code in model training data and the capabilities available in modern browsers. Its CLI approach offers task-time injection of curated recommendations instead of assuming that model weights alone will produce production-ready implementations. As agents take on more frontend work, that gap becomes a concrete source of unnecessary JavaScript, inconsistent accessibility decisions, and old implementation habits.

## Why it matters
For teams using coding agents on web applications, a reusable guidance layer can reduce the need to restate browser-specific constraints in every prompt. Retrieval by guide and codebase scanning make the material more actionable than a general instruction to “use modern web standards”: developers can locate relevant recommendations and identify legacy patterns in an existing project.

The useful operational model is narrow and testable. A team can ask an agent to inspect a particular UI behavior, retrieve guidance for that behavior, propose a change, and then run its usual checks. That is more controllable than expecting a general-purpose model to infer support requirements, accessibility behavior, and preferred platform primitives from a task description alone.

The value still depends on review. A guide can influence an agent’s choices, but it does not guarantee compatibility, accessibility, or performance in a particular application. Teams should use the skills as implementation context and verify generated changes against their supported browsers, design requirements, analytics, and tests. The wider lesson is that teams can shape coding-agent behavior by maintaining focused, machine-usable expertise alongside their codebase.

## Limits and uncertainty
Google Chrome’s repository describes the intended workflow and the scope of the guidance, but this note does not independently measure how much the skills improve generated code, reduce JavaScript size, or catch legacy patterns. The linked evaluation dashboard is part of the project’s own evaluation apparatus, so its results would need separate scrutiny before being treated as independent performance evidence.

Browser compatibility remains feature-specific, and a recommendation that is modern or performant in one target set may need a fallback in another. The project’s stated coverage of 250 features also does not mean every feature is suitable for every production application. Teams retain responsibility for deciding their browser baseline, validating accessibility behavior, reviewing fallbacks, and monitoring the effect of changes after release.

## Skill to build now
Try the CLI against one web project this week: run `npx modern-web-guidance@latest legacy-scan .`, then inspect the findings before asking an agent to modernize anything. For one identified pattern, use `search` or `retrieve` to supply targeted guidance, and compare the proposed change against your supported-browser policy and existing tests.

## Reader takeaway
Modern Web Guidance makes curated web-platform expertise retrievable inside coding-agent workflows. The opportunity is not to trust generated code more automatically. It is to give an agent sharper, current context for a specific web task, then verify the proposed implementation with the same discipline required for any production change.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Google](https://newruntime.com/topics/google.md): Explore the google topic hub.
- topic hub: [Coding Agents](https://newruntime.com/topics/coding-agents.md): Explore the coding-agents topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)