1 · Discover capabilities

One GET tells your agent everything this operator can do — endpoints, auth, protocols, checkout model:

curl https://api.b2ac.ai/v1/agent/capabilities

2 · Search the catalog

Natural-language semantic search over agent-optimized products. Results include price, currency, attributes, and a redirect URL that credits the merchant visit:

curl "https://api.b2ac.ai/v1/agent/search?q=waterproof+hiking+jacket&limit=5"

Can't find something? Record demand — merchants see what agents asked for:

curl -X POST https://api.b2ac.ai/v1/agent/request \
  -H "Content-Type: application/json" \
  -d '{"query": "titanium camping spork shippable to Singapore"}'

3 · Or connect over MCP

The same catalog is exposed as a Model Context Protocol server (streamable HTTP) with search_products, get_product and request_product tools. Add it to any MCP client:

{
  "mcpServers": {
    "b2ac": { "type": "http", "url": "https://api.b2ac.ai/mcp" }
  }
}

Discovery metadata lives at https://api.b2ac.ai/.well-known/mcp.json.

4 · Get a free key for higher limits

Anonymous access works with modest rate limits. A self-serve key is free and instant — send it as X-Agent-Key:

curl -X POST https://api.b2ac.ai/v1/agent/keys \
  -H "Content-Type: application/json" \
  -d '{"agent_name": "my-shopping-agent", "contact": "you@example.com"}'

Feeds & protocols

Per-merchant agent-commerce feeds are published in ACP and UCP (plus Google Merchant Center and Bing formats), with machine-readable incentives and discovery-time effective prices. Deal discovery: GET /v1/agent/deals. UCP discovery per merchant site: /.well-known/ucp.