Writing for the web assistants read.

A TanStack Start and Cloudflare Workers checklist for SSR metadata, canonicals, headers, caching, markdown variants, and scans.

How to verify AI-built websites for crawlability, canonicals, metadata, robots policy, sitemap health, and launch-safe previews.

AI agents are starting to look in /.well-known for service capabilities, security contacts, and OAuth metadata. Here's what to publish and why.

Turn recurring audit fixes into scoped instructions that coding agents can execute reliably.

A field checklist for robots.txt, sitemaps, metadata, and content surfaces that AI crawlers can evaluate.

A repeatable template for evaluating machine-readable surfaces, crawler policy, and remediation priority.

How edge headers, static assets, and Worker-rendered HTML should align for technical SEO.

How to document AI crawler access policies without blocking the discovery paths you actually want indexed.

How MCP cards and public context endpoints can make web tools easier for agents to use.

Where answer, generative, and search optimization overlap, and where technical implementation differs.

How to scope, link, and maintain an llms.txt file that helps AI assistants inspect the right surfaces first.

How llms.txt and XML sitemaps serve different discovery jobs and why production sites often need both.

The security header baseline that protects users while making crawler behavior more predictable.

Why Organization, SoftwareApplication, FAQPage, and Article schema still matter in AI-mediated discovery.

The metadata, canonical, and crawl-quality details that make AI product pages easier to classify.

Three proposals serve three different jobs. Here's what each one does, who reads it, and the minimum viable set most sites should publish.

robots.txt errors don't throw. They get silently ignored and the wrong default applies. Here are the seven bugs that cost real ranking before anyone notices.

Google distrusts sitemaps where every URL claims to have changed today. Here's how to generate an accurate lastmod that actually drives recrawls.

AI systems chunk pages by heading nesting. The five hierarchy bugs that fragment your content and cost citation eligibility.

Most sites ship duplicate metadata for marginal benefit. Here's what each format actually does, where they differ, and the minimal set that covers every share surface.

Pixel-width truncation rules plus the writing pattern that survives Google's rewriter while feeding AI summaries cleanly.

Multimodal models read your images directly now. Bad alt text doesn't just fail screen readers — it contradicts what the AI sees, and that hurts citation.

Trailing slashes, query params, www splits, and mixed protocols. The five canonical bugs that cost ranking and AI citations without an obvious symptom.

Linking out to primary sources signals expertise to AI systems. The "preserve link equity" advice predates retrieval-augmented generation.

AI Overviews and Perplexity quote the first sentence after a heading. Bury the answer and you don't get cited. Here's the structure.

AI Overviews don't just cite the top organic results. Here's the retrieval pipeline they actually use and what that means for your content.

Google killed the FAQ rich result for most sites in 2023, but FAQPage schema still feeds AI Overviews and Perplexity citations. Here's what to ship.

Workers deliver sub-50ms TTFB at the edge when you stay on the cache path. Origin fetches and cold KV reads blow that up. Here's where to look first.

Every documented AI crawler in one table — exact user-agent strings, operator, purpose, and whether it respects robots.txt.

INP replaced FID in March 2024. Here are the current Core Web Vitals thresholds, what they actually measure, and the practical work that moves the needle.

Most LLMs trained on the web pulled from Common Crawl first. Here's what CCBot does, why it's the most leveraged opt-out, and the exact robots.txt syntax.

Apple split AI training away from product indexing in 2024. Here's how the two tokens work and the exact robots.txt to permit one without the other.

Most CSP advice picks one of two failure modes. Here's the report-only-first, hash-or-nonce, strict-dynamic path that ships a real CSP without midnight rollbacks.

Bytespider drives massive crawl volume to feed TikTok and Doubao. The discoverability return for most Western sites is near zero. Here's how to decide.

Preloading bakes HSTS into every major browser, but it's effectively permanent. Here are the requirements, the risks, and when to skip it.

What PerplexityBot crawls, the controversy over robots.txt compliance, and how to express your policy without ambiguity.

AI Overviews and Perplexity weight dateModified for ranking freshness. Here's the distinction, the failure modes, and the right update discipline.

Fonts, images, ads, cookie banners, and async widgets are the usual suspects. Here's how to catch each before it ships.

What GPTBot actually crawls, what blocking costs you in ChatGPT discoverability, and the exact robots.txt syntax for each policy.