// Free & open infrastructure
grow.contact is open infrastructure for AI search visibility. Scan any URL against the Agent-Native Web Standard, connect the MCP server to your assistant, and drop the WordPress plugin into your site. No paywall. No account required. No upsell.
Why it matters: 83% of AI Overview citations come from pages outside the organic top 10, 73% of sites are quietly excluded by fixable technical issues, and AI-referred sessions grew 527% YoY in early 2025.
// What you get, free
Every tool below runs against the same Agent-Native Web Standard. Pick the surface that fits your workflow — the scanner in a browser, the CLI in CI, the plugin on WordPress, or the MCP server inside your AI assistant.
// Tool
Free Scanner
Point it at any URL. Get a 0–100 score, a diff of what agents see, and a prioritised fix list in under 30 seconds.
Open →// Tool
Playground
Try every MCP tool live in the browser. No install, no key — just prompts and results.
Open →// Protocol
MCP Server
Connect Claude, ChatGPT, Cursor, or Codex. Let the agent scan pages and apply fixes itself.
Open →// Tool
Browser Extension
See any page's agent-readability score inline as you browse. Free for Chrome and Firefox.
Open →// Tool
CLI
Fail builds that drop below your threshold. Install with one command, no config required.
Open →// Tool
robots.txt Checker
Verify that AI citation bots (GPTBot, PerplexityBot, ClaudeBot, Google-Extended) can actually reach you.
Open →// Standard
Agent-Native Web Standard
The six-layer open spec. Read it, cite it, fork it — published under CC-BY, versioned like software.
Open →// Docs
Playbooks
Step-by-step guides for shipping structured data, llms.txt, MCP servers, and verifiable claims.
Open →// Data
Leaderboard
Live scores for 2,400+ audited sites across AI, devtools, and agent platforms. Free to browse and query.
Open →· every tool free forever, no card required
// Quick Answers
· refreshed weekly
An agent-native website is a site engineered so AI engines (ChatGPT, Perplexity, Claude, Google AI Overviews) can read, cite, and link to it without rendering JavaScript. It ships semantic HTML, JSON-LD schema, an llms.txt file, an MCP server card, and edge-cached static HTML — the six layers of the open Agent-Native Web Standard.
No. The scanner, the standard docs, the playbooks, the leaderboard, the WordPress plugin, the CLI, the browser extension, the MCP server, and every data drop are free forever. There is no paid tier, no trial, and no billing surface anywhere on this site.
Per the /check scanner dataset (n=2,400+ sites audited as of June 2026), 73% of sites fail at least one of: robots.txt blocking citation bots, JavaScript-only rendering (23% parse success vs 94% for static HTML), missing JSON-LD, or TTFB above the 1–5 second AI-crawler timeout. Each is a small, mechanical fix.
Yes. grow.contact ships an OAuth-protected MCP server at /mcp with tools for scanning URLs, extracting claims, and reading the standard. Any MCP-compatible assistant (Claude, ChatGPT, Cursor, Codex, Codeium) can connect in one click and act on your behalf.
All stats attributed: /check scanner internal benchmark (n=2,400+), 2026.
// Certified Agent-Native
Teams ship the Agent Readability badge on their footers and trust pages. Each one links to a live, re-scored verdict.
// Technical Specifications
Every Grow site ships structured for humans and parseable by LLMs. Here's what actually goes on the wire.
<!-- Semantic, agent-readable markup -->
<main>
<article itemscope itemtype="https://schema.org/Product">
<header>
<h1 itemprop="name">Vector — Embeddings API</h1>
<p itemprop="description">Low-latency vector search…</p>
</header>
<section aria-labelledby="features">
<h2 id="features">Features</h2>
<ul>
<li>p95 < 12ms cold start</li>
<li>OpenAI-compatible endpoint</li>
</ul>
</section>
<footer>
<nav aria-label="Breadcrumb">
<ol>…</ol>
</nav>
</footer>
</article>
</main>// Live demo
Three signals decide whether ChatGPT, Perplexity, Claude or Gemini cite you: structured data, content clarity, and a machine-readable surface. Paste any URL — we check all three.
Paste your URL. We run a real scan and compare it against the live average across every site scanned on grow.contact.
These bars are a real scan summary. The full report on /check breaks down every signal and ships an actionable fix list.
// Questions
✓ FAQPage schema validated · ✓ answer-first structure ·
Is grow.contact really free?
Yes — the scanner, the standard, the playbooks, the leaderboard, the MCP server, the WordPress plugin, the browser extension, the CLI, and every data drop are free forever. No paywall, no trial, no upsell. Bring an account only if you want your scans saved to a dashboard.
What does grow.contact actually do?
grow.contact is open infrastructure for AI search visibility. Point the scanner at any URL and get a 0–100 score against the Agent-Native Web Standard, a diff of what ChatGPT, Perplexity, Claude and Google AI Overviews see when they crawl the page, and a checklist of concrete fixes ordered by impact.
What is the Agent-Native Web Standard?
A six-layer open specification for making a website legible to AI engines: semantic HTML, JSON-LD schema, llms.txt, an MCP server card, edge-cached static HTML, and a verifiability layer. Published under CC-BY, versioned like software, forkable on GitHub.
How is the scanner different from PageSpeed or Lighthouse?
Lighthouse checks whether a browser can render your page. grow.contact checks whether an agent can cite it — robots.txt permissions for AI crawlers, JSON-LD validity, llms.txt presence, TTFB against the 1–5 second AI-crawler timeout, JavaScript-only content, verifiable claims, and MCP discoverability.
Can I self-host or fork it?
Yes. The scanner, the standard docs, the WordPress plugin, the CLI, and the MCP server are open source. You can run the full stack on your own infrastructure or fork the standard and publish a variant. Attribution appreciated, not required.
How do I plug this into my site?
Three paths, pick one: install the WordPress plugin for auto-fixes on save, run the CLI in CI to fail builds that drop below your threshold, or connect the MCP server to Claude/ChatGPT/Cursor and let the agent audit and fix pages itself.
Do you collect my data?
The scanner fetches the URL you submit and stores the score plus a summary. It never stores full page bodies, never runs JavaScript against your users, and never sells anything. The database schema and retention policy are public.
Who maintains it?
Started and maintained by Gudmundur Eyberg Kristjansson in Reykjavík, with pull requests and issue reports welcome from anyone who audits, cites, or ships against the standard.
// Go deeper
Every claim on this site is documented on its own page — the standard, the crawler matrix, the scoring methodology, the citation research. Read, cite, and fork whatever you need.
// Standard
Agent-Native Web Standard
The six-layer open spec. CC-BY, versioned, forkable.
Read →// Matrix
8 AI Systems, 8 Crawlers
How ChatGPT, Perplexity, Claude, Gemini, Copilot and others crawl and cite.
Read →// Guide
GEO vs SEO
What generative engine optimization is, and what moves AI citations.
Read →// Guide
AEO vs GEO vs SEO
Side-by-side comparison of the three optimisation frameworks.
Read →// Guide
llms.txt Spec
The complete specification with examples and validation rules.
Read →// Method
How the Score Works
Five weighted signals — semantic HTML, JSON-LD, llms.txt, citability, speed.
Read →// Data
Live Leaderboard
Scores for 2,400+ audited sites across AI, devtools, and agent platforms.
Read →// Research
Reports & Data Drops
Quarterly report, monthly citation index, playbooks, glossary — every artifact in one hub.
Read →// Journal
Journal
Field notes on AI search visibility, updates to the standard, and post-mortems.
Read →