Skip to main content

Manifesto · April 2026

One URL every AI can read.

Three-minute read. Written by the team building more.md and the open Entity Engagement Protocol.

The web was not built for this

The web was built for browsers driven by humans. Every primitive assumes a person will do the clicking. The URL bar. The forms. The cookies. The Stripe checkout page. The email signup.

That assumption broke quietly, then loudly. By 2026 a growing slice of every HTTP request comes from an agent. Cursor reads your repo. Claude reads your docs. Perplexity cites your homepage. Your CFO's internal copilot reads your pricing page on its way to a budget decision.

None of those agents have a browser. They cannot click. They cannot fill in forms. They scrape. They hallucinate. They quote your old prices because the new ones live behind a SPA.

The three flows that are silently broken

  1. Read and cite without scraping. An agent can fetch your homepage. It cannot ask for a typed self-description. A stable handle. A canonical URL for each section. So it scrapes whatever HTML you happened to render today. The answer drifts.
  2. Pay per read without checkout pages. Stripe checkout in a headless browser is not a payment flow. It is a browser theatre that breaks the moment your CSS changes. Agents need a wire-level payment primitive.
  3. Subscribe without email. Webhooks exist. Every service rolls its own auth and signature scheme. Agents need one verifier that works across endpoints they have never seen before.
We are not building an identity layer. We are not building a wallet. We are building the boring contract that makes those three flows boring.

What more.md is

more.md is the reference implementation of the open Entity Engagement Protocol (EEP). One canonical URL per entity. Three serializations on the same URL. Markdown. JSON. TOON. Picked by an Accept header. A typed envelope. A search endpoint. A payment gate. A real-time channel. A commerce state machine. All on plain HTTP. All MIT-licensed at the core.

If you have ever shipped a REST API, you know how to ship more.md. If you have ever written an OpenAPI spec, you know how to read EEP. There is no SDK lock-in. There is no wallet to install. There is one specification. One set of schemas. A growing set of compliant servers.

What we are deliberately not building

  • An identity layer. Use the DIDs you already have. EEP rides on them; it does not replace them.
  • A wallet. Use the wallet you already have. EEP reads receipts; it does not custody funds.
  • A new auth protocol. Bearer tokens. OAuth. Signed requests. All of them already work. EEP defines the handshake, not the credential.
  • An “agent registry”. Anyone can run a more.md instance. Federation is a property of the protocol, not a business model.

Why open core

EEP is the contract. Every line of the spec, the schemas, the compliance CLI, the SDKs plus the reference server is open and lives at github.com/eep-dev/eep. You can self-host every byte of it.

more.md is the hosted reference. We make money by running the implementation well. Fast pulse. Durable webhooks. Hosted commerce. Not by holding the protocol hostage. If a competitor implements EEP better, that is the protocol working as designed.

What you can do today

  1. Claim a free profile at /register. Five minutes. No credit card.
  2. Drop the Agent Starter snippets into your project.
  3. Scan any URL at /scan to see how AI-readable it is today.
  4. Read the blog. We ship one technical post plus one playbook every week.

We will ship boring infrastructure for the next decade so the agentic web does not get re-invented as a walled garden. If that resonates, come build with us.