Integrations
ANML is designed to work alongside the protocols and frameworks powering the agentic web. It doesn't replace them — it gives agents the semantic context they need to use them intelligently.
Where ANML fits
The agentic web is built on layers. Transport protocols define how agents call APIs. Capability schemas define what data structures to send. But neither layer tells an agent why it should act, what the service means, or how to behave when representing it to a user.
ANML is the comprehension layer — a universal duckument format that any conforming agent can read to understand a service without bespoke integration.
| Layer | Answers | Examples |
|---|---|---|
| Transport | How do I call this? | REST, MCP, A2A, gRPC |
| Capability | What data do I send? | OpenAPI, JSON Schema, UCP schemas |
| Comprehension (ANML) | What does this mean? How should I behave? | ANML duckuments |
Universal Commerce Protocol (UCP)
UCP defines the wire-level contract for agentic commerce: checkout sessions, payment handlers, fulfillment, capability negotiation. It tells an agent exactly which endpoints to call and what JSON to send.
What UCP doesn't provide is the semantic framingan agent needs to act intelligently within that contract. That's where ANML comes in.
What ANML adds to UCP
Workflow visibility
The <flow>element maps to UCP's multi-step lifecycle (cart → checkout → payment → fulfillment), giving agents an explicit view of where they are.
Privacy governance
<constraints> declares which fields require explicit consent — addressing compliance concerns UCP leaves to implementers.
Contextual knowledge
<knowledge> communicates payment methods, shipping policies, and fraud-prevention signals alongside the protocol contract.
Behavioral guidance
<persona> tells the agent how to present information to the user — tone, language, confirmation requirements — with no equivalent in API specs.
Example: UCP checkout with ANML context
A merchant publishes an ANML duckument at /.well-known/anml that wraps their UCP checkout flow with semantic context:
{
"anml": "1.0",
"ttl": 300,
"head": {
"title": "Checkout — Acme Electronics",
"meta": [
{ "name": "protocol", "value": "ucp" },
{ "name": "protocol-version", "value": "2026-04-08" }
]
},
"constraints": {
"disclosure": [
{ "field": "payment-credential", "requires": "explicit-consent" },
{ "field": "shipping-address", "requires": "explicit-consent" },
{ "field": "buyer-ip", "requires": "none" }
]
},
"state": {
"context": { "step": "checkout" },
"flow": {
"step": [
{ "id": "cart", "label": "Cart", "status": "completed" },
{ "id": "checkout", "label": "Checkout", "status": "current", "action": "create-session", "required": true },
{ "id": "payment", "label": "Payment", "status": "pending", "action": "complete-payment", "required": true },
{ "id": "confirm", "label": "Confirmed", "status": "pending" }
]
}
},
"interact": {
"action": [
{ "id": "create-session", "method": "POST", "endpoint": "/ucp/checkout-sessions", "enctype": "application/json", "auth": "required" },
{ "id": "complete-payment", "method": "POST", "endpoint": "/ucp/checkout-sessions/{id}/complete", "enctype": "application/json", "auth": "required", "confirm": true }
]
},
"knowledge": {
"inform": [
{ "confidentiality": "public", "ttl": 3600, "content": "Accepted: Visa, Mastercard, Apple Pay, Google Pay. Free shipping over $75. 30-day returns." }
],
"ask": [
{ "field": "buyer-ip", "action": "create-session", "required": true, "purpose": "fraud-prevention" },
{ "field": "shipping-address", "action": "create-session", "required": true, "purpose": "shipping-calculation" },
{ "field": "payment-credential", "action": "complete-payment", "required": true, "purpose": "payment" }
]
},
"persona": {
"tone": { "value": "helpful" },
"instructions": "Show itemized totals including tax. Confirm the final amount before requesting payment. Mention the return policy."
},
"body": { "content": "4K Monitor x1 — $599.00 | USB-C Cable x2 — $29.98 | Subtotal: $628.98" }
}Use Cases
Agent-powered shopping
A shopping agent discovers a merchant's ANML duckument, understands the checkout flow, knows what consent is needed for payment, and presents options to the user in the merchant's preferred tone — all without custom integration per merchant.
Travel booking orchestration
A travel agent navigates multi-step booking flows (search → select → pay → confirm) using the flow element for state awareness. The knowledge section provides real-time availability and pricing context. Constraints ensure passport data is only shared with explicit consent.
IoT device interaction
A smart home agent reads an ANML duckument from a thermostat service to understand available actions (set temperature, schedule, read history), what data it needs (location for weather-aware scheduling), and how to present controls to the user.
Healthcare appointment scheduling
A patient's agent discovers a clinic's ANML duckument. Constraints enforce that medical history requires explicit consent. The flow tracks the booking lifecycle. Knowledge informs the agent of available slots and insurance requirements.
Enterprise SaaS onboarding
An IT admin's agent reads ANML duckuments from SaaS providers to understand provisioning workflows, required configuration data, and compliance constraints — enabling automated multi-vendor setup without per-vendor scripting.
Financial services
A personal finance agent uses ANML to understand bank and brokerage services: what actions are available (transfer, trade, query), what disclosure rules apply to account numbers and SSNs, and what confirmation is required before executing transactions.
Compatible protocols
ANML is protocol-agnostic. Any system that defines structured agent interactions can benefit from an ANML comprehension layer. Here are some natural pairings:
| Protocol | What it provides | What ANML adds |
|---|---|---|
| UCP | Commerce endpoints, capability negotiation, payment flows | Workflow context, disclosure rules, persona, knowledge |
| MCP | Tool definitions, resource access, prompt templates | Service identity, behavioral guidance, privacy constraints |
| A2A | Agent-to-agent communication, task delegation | Semantic context for delegated tasks, trust boundaries |
| OpenAPI | REST endpoint schemas, request/response definitions | Workflow state, knowledge exchange, usage rights |
Build with ANML
If you're building an agent platform, commerce service, or any system that autonomous agents interact with, ANML gives your users' agents the context they need to act intelligently — without custom integration work.