What it is, how to implement it, how to verify it. Services AI agent readiness audit WebMCP Implementation Agentic SEO architecture Structured data and schemas API optimization and data layer for AI Knowledge graph and entity modeling WebMCP migration and refactoring WebMCP consulting for companies and agencies WebMCP for online stores Origin trial and technical compliance WebMCP maintenance and monitoring See all services Blog Leaderboard RO EN
Real Chrome execution

WebMCP, the standard that lets AI agents use your site

Browser AI agents do not crawl a page the way a search bot does. They call the tools your site registers. We load your site in a real Chrome, intercept the modelContext API before the first page script runs, and show you exactly which tools end up available. Deferred, bundled and base64 encoded scripts included.

WebMCP.ro
Takes about 30 seconds and gives you a shareable report. Limit is 8 checks per hour.

Runs in a real browser

A text search through HTML cannot see a registerTool() call inside a deferred or base64 encoded bundle. Our Chrome opens the page, scrolls it, simulates interaction and records every tool that genuinely registers.

Checks that follow the spec

We test only what the official Chrome documentation defines: toolname, tooldescription, toolparamdescription, the modelContext API and the origin trial token. No invented attributes.

Scoring we can defend

Community conventions such as the /.well-known/webmcp manifest are labelled unofficial and scored separately, as a bonus. Anything that cannot be checked remotely is stated as such rather than failed.

Why the other WebMCP checkers get it wrong

  • WebMCP tools only exist inside a real browser tab. A scanner that does not execute JavaScript reports zero tools for sites that work perfectly.
  • Performance plugins defer and base64 encode scripts, so searching the page source for registerTool finds nothing even when the tools work.
  • There is no manifest, data-toolname or toolaction in the official spec. A checker that fails you on invented attributes is not checking anything real.
  • While WebMCP is in origin trial, a valid token is exactly what makes tools work in stable Chrome. Almost nobody checks it. We decode it and confirm its origin and expiry.
What we do not claim to check: whether an agent can carry your tools through to completion, human confirmation flows, or the quality of the results. Those need a real agent session, not a scan. Anyone scoring them remotely is guessing.

Recent checks

WebsiteScoreGradeWhen
www.domro.ro 92/100 A 1 hour ago
www.webmcp.ro 100/100 A+ 1 hour ago
www.webmcp.ro 100/100 A+ 2 hours ago
www.webmcp.ro 86/100 A 17 hours ago
www.wpsecurity.ro 100/100 A+ 22 hours ago
asigurari-ssl.ro 100/100 A+ 1 day ago

WebMCP services

🔍 AI agent readiness audit

We analyze your site from an AI agent's perspective and deliver a report ranked by impact, with causes and solutions for each find...

🔧 WebMCP Implementation

We write and deploy complete WebMCP tools, with manifest, origin trial and final verification in a real Chrome browser, on your ex...

🧭 Agentic SEO architecture

We restructure your site's architecture so that AI agents can understand and use the information, not just so search engines can i...

🏷️ Structured data and schemas

We implement and verify correct schema.org structured data, consistent with the content displayed, laying the groundwork for WebMC...

🔌 API optimization and data layer for AI

We prepare and optimize the API behind your website so that WebMCP tools receive stable, accurate and timely data.

🕸️ Knowledge graph and entity modeling

We define your company's business entities and the relationships between them, so that AI models and AI agents understand you corr...

See all services »

Want WebMCP working on your site?

We implement WebMCP end to end: imperative tools, valid schemas, the origin trial token and the discovery manifest. We work on the platform you already run, and the result is verifiable right here.

See our services

Latest articles

What Lighthouse's "WebMCP form coverage" means and how to fix it

What Lighthouse's "WebMCP form coverage" means and how to fix it

A practical guide to the WebMCP form coverage check in Lighthouse: which attributes actually exist and how to...

3 August 2026
How to check whether your WebMCP tools work in production

How to check whether your WebMCP tools work in production

A guide to checking WebMCP in production: what gets tested, how the test is repeated, and what frequently brea...

23 July 2026
Why static WebMCP checkers give wrong results

Why static WebMCP checkers give wrong results

We explain technically why WebMCP checkers based on HTML text search produce wrong results, and what a correct...

21 July 2026

All articles »