Version 0.1 · Public Draft
An open protocol for AI-readable commerce catalogs.
The Enhanced Catalog Layer of agentic commerce. AIRCP defines what catalogs publish — both buyer-agnostic product attributes (Tier 1) and buyer-relevant attributes (Tier 2) designed to match against a buyer profile. Composes with UCP, AP2, A2A, and MCP. The Relevancy Reasoning Layer that matches Tier 2 attributes against a specific buyer is intentionally outside the protocol — where AI agents compete. MIT licensed.
The Stack
AIRCP composes with the protocol stack that already exists.
Four open protocols define agentic commerce today. MCP exposes tools to agents. A2A connects agents to agents. AP2 authorizes agent-initiated payments. UCP handles checkout and order management. AIRCP completes the stack with the Enhanced Catalog Layer. A fifth layer — the Relevancy Reasoning Layer, where agents match catalog data against a specific buyer — is intentionally out of protocol scope, where AI agents compete.
| Layer | Protocol | What it defines | Status |
|---|---|---|---|
| Tool exposure | MCP | How AI tools expose themselves to agents | Widely adopted |
| Agent communication | A2A | How agents talk to other agents | Adopting |
| Payment authorization | AP2 | Verifiable user consent for agent payments via Intent / Cart / Payment Mandates | Adopting (Apache 2.0) |
| Commerce transaction | UCP | Unified checkout sessions, identity linking, order management | Adopting (Apache 2.0) |
| Enhanced Catalog | AIRCP | Tier 1 enriched product attributes and Tier 2 buyer-relevant attributes | Public draft (MIT) |
| Relevancy Reasoning | (intentionally out of protocol scope) | How an agent matches Tier 2 attributes against a specific buyer's profile | Agent-side, proprietary |
AIRCP defers to UCP for transactions and to AP2 for payment authorization. The Relevancy Reasoning Layer — how a specific buyer's profile shapes the recommendation — is intentionally not part of the protocol. That layer is where AI agents differentiate. AIRCP standardizes the data inputs to that layer without prescribing how it operates.
Architecture
Two tiers in the catalog. Two layers in the stack.
AIRCP defines two tiers of attributes. Both are open. Both are published in the catalog. They serve different purposes.
Enhanced Product Attributes
UCP base data plus enriched descriptive and relational attributes. Material, color, weight, style classification, image projection, context suitability, product-to-product compatibility, substitution reasoning, weather and seasonal suitability, aesthetic compatibility. Everything an agent needs to know about a product, regardless of who is buying.
Relevancy Attributes
Interpretive attributes designed to be matched against a buyer's profile. Fit profile, persona compatibility, occasion suitability. Published openly in the catalog. Interpretively useful only when matched against a buyer model — and the buyer model is intentionally outside the protocol.
The Relevancy Reasoning Layer
AIRCP does not define how an agent matches Tier 2 attributes against a specific buyer. That layer — buyer modeling, taste reasoning, the matching engine that turns catalog data into personalized recommendations — is the Relevancy Reasoning Layer. It is intentionally outside the protocol because it is where AI agents differentiate from each other. Standardizing it would erase that differentiation. AIRCP makes the catalog data common. Agents compete on what they do with it.
Why AIRCP
The commerce internet was built for browsers, not agents.
Discovery
No standard way for an agent to find which catalogs exist or how to query them. Agents resort to web scraping, search APIs, or per-merchant integrations.
Interpretation
Product data is inconsistent across merchants. Attribute names differ, types are unspecified, important context like materials and sizing is buried in unstructured text or absent entirely.
Transaction
Even when an agent finds the right product, completing the purchase requires merchant-specific integration. No standard for add-to-cart, checkout, or purchase confirmation.
What AIRCP does
A thin, semantic protocol any catalog can implement.
Discovery
A well-known URL pattern. Any agent can find an AIRCP-compatible catalog by checking /.well-known/aircp.json on any domain.
Structured catalog
A JSON representation of products with two tiers of attributes: Core (mandatory, deterministic) and Enhanced (optional, semantic).
Transaction hooks
Standard methods for agents to query, add to cart, check out, and receive purchase confirmations.
Trust layer
Merchant verification so agents know which catalogs are legitimate. Optional for v0.1, recommended for production.
Organizing the inventories of the world into AIRCP.
AIRCP is an open standard. Implementations are welcome from anyone.
Get Started
Implement AIRCP in an afternoon.
The Core tier of AIRCP is small enough to be implemented by any commerce site in a few hours. The full specification is publicly available, MIT licensed, and free to implement.
AIRCP was authored by the Shoop team and is maintained publicly. We welcome contributions, implementations, and protocol critique.