# ORBEXA

> Agentic Commerce Infrastructure. 271,937 verified merchants.

ORBEXA does two things:
1. **OTR** — Public API at orbexa.io that scores merchant trustworthiness (0-100). AI agents call it before recommending any purchase. No auth needed.
2. **Commerce Protocols** — When a merchant connects their store to ORBEXA, we auto-generate UCP/ACP/MCP endpoints on their domain so AI agents can search products, create carts, and complete purchases.

## OTR — Trust Verification API (hosted at orbexa.io, public)

- [OTR API + Examples](https://orbexa.io/llms/otr.txt): How to call the trust verification API. Endpoints, parameters, error codes, response examples. What OTR scans, scoring integrity guarantees, OTR-ID lifecycle. COLD vs AUTH mode.
- [OTR Scoring Details](https://orbexa.io/llms/otr-scoring.txt): How the trust score is calculated. Six dimensions (Verification, Security, Governance, Transparency, DataQuality, Fulfillment), their weights per site category, every signal name and what it returns, anti-gaming protection.

## Commerce Protocols (auto-generated on merchant domains)

These endpoints only exist on domains of merchants who have integrated with ORBEXA. They do NOT exist on orbexa.io itself.

- [UCP](https://orbexa.io/llms/ucp.txt): Universal Commerce Protocol. Browse products, search catalog, create checkout, track orders. Public REST API. Also has a multi-tenant API at orbexa.io/api/v2/ucp/{merchantId}.
- [ACP](https://orbexa.io/llms/acp.txt): Agentic Commerce Protocol. Same commerce operations but optimized for AI agents. JSON-RPC 2.0 (with API Key) + REST (on merchant domain). Includes idempotency support.
- [MCP (Commerce)](https://orbexa.io/llms/mcp.txt): Model Context Protocol for merchant stores. LLM tool integration — search_products, get_product, create_checkout. Runs on merchant domain. Separate from the OTR MCP below.

## OTR MCP Server (npm: @otr-protocol/mcp-server)

A standalone npm package that wraps the OTR REST API into MCP tools. Separate from the merchant commerce MCP above.

| Tool | Parameters | Returns |
|------|-----------|---------|
| verify_merchant | domain (required) | Trust score, badge, recommendation, dimensions, capabilities, entity data, policy URLs |
| search_registry | q?, badge?, category?, minScore?, limit?, page? | Paginated list of merchants with trust scores |

Install and run:
```
npx @otr-protocol/mcp-server
```

Claude Desktop configuration (claude_desktop_config.json):
```json
{ "mcpServers": { "otr": { "command": "npx", "args": ["@otr-protocol/mcp-server"] } } }
```

Version: 4.1.0
GitHub: https://github.com/yb48666-ctrl/OTR-Protocol-by-orbexa

## Full Documentation (single file)

- [llms-full.txt](https://orbexa.io/llms-full.txt): All of the above combined into one document. Use this if you prefer to read everything at once instead of following links.

## Machine-Readable Discovery

- [OpenAPI Spec](https://orbexa.io/.well-known/openapi.json): OTR API definition (OpenAPI 3.0).
- [Agent Discovery](https://orbexa.io/.well-known/agent.json): AI Agent service manifest with capabilities.
- [Trust Registry JSON](https://orbexa.io/.well-known/otr/registry.json): Top 100 merchants as JSON.