There is a new scoreboard for websites and most of the internet is failing it. Agent-readiness scanners test whether an autonomous agent can find your site, read it without burning tokens on your markup, use your APIs and pay you — all without a human ever rendering a page. Twenty-one checks. We scanned a hundred of the web’s most-used sites and the median scored 8 of 21 — not one earned an A or a B, and the best managed 15.
moltlinestudio.com scores 100 out of 100 on isitagentready.com, an independent scanner we do not control, with every optional commerce check green too. It scores 21 of 21 on our own checker. We took it there ourselves, the hard way, and wrote down every undocumented shape the scanners actually demand. That is the work we are selling.
Score your own domain free Twenty-one checks, about thirty seconds, no signup. You will see exactly what fails before you talk to us.
The web's next visitors do not have eyes. When someone tells an agent “find me a vendor that does X and set it up”, the agent does not browse. It reads machine-readable surfaces, follows discovery records, and decides without ever seeing your hero section.
An agent shortlisting vendors reads whatever it can parse. If a competitor publishes a service catalogue, pricing and an agent-readable access guide and you publish a JavaScript app, you are not in the comparison at all — and nobody tells you that you lost.
Every one of these checks is small on its own. Doing them before the standards calcify is a week of work. Doing them after your competitors, under pressure, with a marketing deadline attached, is a project.
This is not an opinion about content quality. There is a public scanner, it returns a number, and the number moves. You can verify the outcome yourself in thirty seconds without trusting us.
Twenty-one checks across five groups. Most agencies have not heard of any of them.
| Group | What gets built |
|---|---|
| Discoverability | robots.txt with AI and Content-Signal directives, sitemap, RFC 8288 Link headers, and DNS-based discovery records signed with DNSSEC |
| Understanding | llms.txt, markdown content negotiation so agents read you at a fraction of the token cost, and structured data validated to zero errors |
| Protocol surfaces | RFC 9727 api-catalog, MCP server card, agent-skills index with digests, A2A agent card, WebMCP registration, and an auth guide agents can actually follow |
| Identity & trust | Web Bot Auth signing keys, OAuth discovery and protected-resource metadata shaped the way validators expect rather than the way the examples show |
| Commerce | x402 payment challenges, plus ACP, UCP, MPP and AP2 discovery so an agent can transact with you without a human in the loop |
Fixed price, fixed scope, stated up front. No day rates, and no scoping call before you know what it costs. Pick a package below and it will assemble the email for you.
The 21-check scan itself costs nothing and always will — run it on the free checker right now, and you will get your score, every failure, and the fix for each one. If your site is simple and you have an engineer with a spare afternoon, that may genuinely be all you need, and we would rather you did that than paid us for a PDF of a free tool. What is below is for the cases where the scan is the easy part: multiple properties, a stack that fights you, a team that needs the plan written down, or a deadline.
For teams who have engineers and just need to know precisely what to build, in what order, and what it is worth.
These are young standards. A domain that passes today drifts. This is the difference between a project and a property.
Choose a package above, add your domain, and this will open a prefilled email. Nothing is sent from this page and nothing is stored — it just writes the message for you.
Tip: run the free scan first and paste your score in — it makes the first reply useful instead of a questionnaire.
Multi-property rollouts, platforms that need every customer site to be agent-ready, or teams who want the capability transferred to them rather than delivered: from $18,000, scoped properly before anyone commits. Email us with what you are working with.
Not a claim in a deck. Scan our domain with a tool we do not control and read the number. We are one of a very small number of sites at that level.
Every one independently MCPize Verified A across seven dimensions, and A / 100% on Glama's quality rubric. We run this stack, we do not just talk about it.
The protocol reference gives away the undocumented shapes the scanners demand. People who hide their method are usually selling the method. We are selling the execution.
The ones that come up before anyone signs anything.
Scanners test whether an autonomous AI agent can discover your site, read it efficiently, use your APIs and pay you, without a human rendering a page. Sites that pass get recommended, cited and bought from by agents. Sites that fail are invisible to them. It is the same argument as mobile-friendliness in 2013, and it will look just as obvious in hindsight.
You verify it with a scanner we do not control. Run your domain through isitagentready.com before and after. The score is the deliverable, so you never have to take our word for anything.
On an implementation engagement the second half is not payable. We have done this end to end at maximum difficulty, so we are comfortable carrying that risk instead of asking you to.
For the audit, no — everything is measured from outside. For implementation we need somewhere to put files and headers, usually a pull request against your web server config. We never need database access, customer data, or production credentials.
Audit: five business days. Implementation: two weeks on a normal stack. Unusual infrastructure or a slow change-approval process extends it, and we will tell you that before you pay rather than after.
Yes, and some of you should. The reference is free and complete enough to work from. You are paying us for the time it saves and for not rediscovering which undocumented field each validator wants.
Run your domain through isitagentready.com and send us the result. If the number is already high, we will say so and you will have spent nothing. If it is not, we will tell you what the gap is worth fixing and what it would cost, before you commit to anything.
Email us your domain and score
support@moltlinestudio.com — a human replies, usually within a day. Invoiced in USD, card or bank transfer, 50/50 on implementation engagements.