Agora
The internet's missing commerce layer.
Built for AI agents. Open for everyone.
The internet was built for human browsers. AI agents need to discover, search, and purchase from stores programmatically - but there's no standard interface for that. Agora is the open protocol that makes every store agent-ready.
Three layers. One protocol.
Agora connects AI agents to e-commerce stores through an open protocol. Stores publish a manifest describing their capabilities. Agents discover stores through a public registry. Purchases happen through a consumer-approved transaction layer.
Like robots.txt tells crawlers what to index, agora.json tells agents what a store sells and how to buy it. A simple JSON manifest at /.well-known/agora.json declares the store's products endpoint, search capabilities, authentication, and rate limits.
The public registry indexes every protocol-compliant store. Agents search across all stores simultaneously - one query returns products from 52 stores, ranked by relevance. Cross-store matching finds the same product at different prices. Trust scores surface the most reliable stores.
Agents build carts and request checkout. Consumers approve each purchase either inline ("Buy these boots for $89?") or via a secure approval link. Tokenized payments mean Agora never stores card numbers. Stores receive order notifications via webhooks.
Build agents that shop
Your agent shouldn't need to scrape HTML, reverse-engineer checkout flows, or handle a different API for every store. Agora gives you one unified interface to search 22,000+ products, compare prices across stores, and complete purchases - all through a single SDK.
Three integration paths: a TypeScript SDK with built-in caching, an MCP server for Claude and ChatGPT, or direct REST API calls. Full OpenAPI spec at /openapi.json and an interactive playground to test every endpoint before writing a line of code.
import { Agora } from 'agora-sdk'
const agora = new Agora({ apiKey: 'ak_your_key' })
const results = await agora.search('hiking boots under $100'){
"mcpServers": {
"agora": {
"url": "https://agora-ecru-chi.vercel.app/mcp"
}
}
}Make your store agent-ready
AI agents are the next sales channel. When a consumer tells their AI assistant to "find me hiking boots under $100," your store should be in those results. Agora makes that happen - instantly for Shopify stores, or through a simple protocol spec for custom platforms.
Stores that join the protocol get listed in a public registry that every agent on the network can query. You get analytics showing how agents interact with your products, a trust score that boosts your visibility, and webhook notifications for every search, view, and purchase.
For Shopify stores, it's a single API call - no code changes, no app installs, no configuration. We generate your protocol manifest, proxy your product feed in the standard format, and register you in the public registry. For custom platforms, implement two endpoints and validate with our CLI tool.
curl -X POST https://agora-ecru-chi.vercel.app/v1/adapter/shopify \
-d '{"url": "https://your-store.com"}'npx @agora/validator https://yourstore.comThe infrastructure layer
for agent commerce
The internet is being rebuilt for AI agents. Today, agents can read, write, and reason - but they can't buy. There's no standard way for an AI to discover what a store sells, compare prices, or complete a purchase. Agora is that standard.
Like Stripe built the payment rails for the internet, Agora is building the commerce rails for the agent era. Stores implement a simple protocol (agora.json at /.well-known/), agents discover them through a public registry, and transactions flow through a consumer-approved checkout layer.
The protocol is live. The registry is public. The transaction layer works end-to-end. 52 stores and 22,562 products are already indexed. The Shopify adapter means any of 4 million+ merchants can join with a single API call. The question isn't whether agent commerce will happen - it's who builds the infrastructure.
This isn't a whitepaper
Everything described on this page is deployed and running in production. Try any endpoint right now.