A2WF — Agent-to-Web Framework – The open standard for AI agent governance on the web

Open Standard v1.0

A2WF defines what AI agents can find and do on your website, and what's strictly forbidden.

The AI Agent gets detailed information and strict rules from siteai.json. No more guessing, no more gray zones!

Without A2WF vs. With A2WF

The difference between hoping AI agents behave and having proof when they don't.

❌ Without A2WF	✅ With A2WF
Scraping is unwanted	Scraping is provably forbidden
Auto account creation just happens	Account creation explicitly prohibited & documented
Agents log in uncontrolled	Login rules machine-readable & dated
Bulk data extraction, no proof possible	Bulk extraction dated & forbidden in policy
Price monitoring by competitors	Price scraping policy-violating & provable
In court: "We didn't want this"	In court: "Here is the dated, machine-readable document"

⚖️ Why A2WF Matters in Court

Without a machine-readable policy, you cannot prove an AI agent violated your rules. Your Terms of Service are written in natural language on a subpage. No agent reads them, no judge can prove the agent could have read them.

With A2WF you have:

A dated, machine-readable document at your website root
That every agent can automatically read (and should)
With explicit prohibitions in a standardized structure
Retrievable via documented discovery mechanisms
Comparable to robots.txt but legally actionable

The critical difference:

Today, your only option is robots.txt which says "please don't crawl." It's a suggestion. Agents ignore it.

A2WF says: "This behavior is forbidden under our documented policy, in these jurisdictions, and here are the legal grounds."

That turns a suggestion into evidence.

What's Happening to Websites Right Now

AI agents are visiting your website every day. They scrape your content, try to log in, create fake accounts, solve your CAPTCHAs, extract your pricing, and harvest your data for model training. They don't read your Terms of Service. They don't ask for permission. And right now, you have no way to prove they broke your rules.

What A2WF Protects You Against

Real threats that websites face today and how A2WF gives you provable, enforceable protection.

🚨

Content Scraping

AI agents extract your entire product catalog, pricing, and content in minutes. Without A2WF: legal gray zone. With A2WF: provably forbidden.

🤖

Automatic Account Creation

Bots create hundreds of fake accounts on your platform. You block IPs, but you can't prove it was forbidden. A2WF makes it explicit: "createAccount": false.

🔐

Automated Logins

Agents log in automatically and act on behalf of users. Your terms say "forbidden" but no agent reads terms. A2WF they do.

📊

Bulk Database Queries

Your search function is abused as an API, thousands of queries per hour. Rate limits in A2WF are machine-readable and enforceable.

💰

Price Monitoring

Competitors let agents monitor your prices in real time. "priceMonitoring": false, one entry that counts in a dispute.

🧠

Training Data Harvesting

LLM providers scrape your content for training data. "trainingDataUsage": false, clear, dated, provable.

One file. Complete control.

Place an A2WF policy file on your web server. AI agents read it before interacting, a clear, enforceable rulebook for the AI era.

Get Started

{
  "@context": "https://schema.org",
  "specVersion": "1.0",
  "identity": {
    "@type": "WebSite",
    "name": "Example Store",
    "category": "e-commerce"
  },
  "permissions": {
    "read": {
      "productCatalog": { "allowed": true },
      "pricing": { "allowed": true }
    },
    "action": {
      "search": { "allowed": true },
      "checkout": {
        "allowed": true,
        "humanVerification": true
      },
      "createAccount": { "allowed": false },
      "submitReview": { "allowed": false }
    },
    "data": {
      "customerRecords": { "allowed": false },
      "paymentInfo": { "allowed": false }
    }
  },
  "scraping": {
    "bulkDataExtraction": false,
    "priceMonitoring": false,
    "trainingDataUsage": false
  }
}

Built for the AI era. Not for crawlers.

Modern AI agents don't just crawl, they interact, transact, and extract. A2WF is the governance layer the web has been missing.

❌

Old approach: Binary Allow/Deny

robots.txt can only say "crawl this" or "don't crawl this". No concept of actions, authentication, or rate limits.

✅

A2WF: Granular Permissions

Define what agents can read, what actions they can take, what data they can access, and under what conditions.

❌

Old approach: No Identity

Websites can't describe themselves in a structured way that agents understand.

✅

A2WF: Rich Metadata

Name, category, language, contact, business hours, everything an agent needs to interact intelligently.