The Website Specification

295 points · 118 comments on HN · read original →

Platform-agnostic checklist of 128 technical features websites should implement, organized across ten categories from SEO to accessibility.

The Website Specification is a reference document mapping ten categories of web features to established standards: Foundations (HTML/head basics), SEO (robots.txt, sitemaps, structured data), Accessibility (WCAG rules), Security (headers, policies), Well-Known URIs (/.well-known/ paths), Agent Readiness (AI legibility), Performance (Core Web Vitals, caching), Privacy (consent signals), Resilience (error handling, redirects), and Internationalization (language, locale).

Each topic links to source standards from WHATWG, W3C, IETF, WCAG, and MDN rather than opinions. The spec is implementation-agnostic across WordPress, Django, Next.js, Hugo, and plain HTML. The full specification is available as an MCP server and through /llms.txt endpoints on each page. Users audit sites against a checklist, learn implementation details, and can submit PRs with sources.

What HN community is saying

Commenters split between appreciation for the reference and skepticism about scope. Top-ranked comments note the existence of validation scripts and praise discovery of /.well-known/ URIs. Criticism centers on three points: whether Agent Readiness features will age well if AI needs change (with one commenter noting they are likely "temporary measures" for token costs), whether sites will implement machine-readable features that might enable deception, and whether the spec conflates actual requirements (title tags, HTTPS) with opinion-based best practices. One experienced developer praised the checklist for picking up overlooked basics after 30 years of work. A recurring complaint emerged that modern web bloat contradicts the spec's simplicity ideals.