OnwardTicket Developer Documentation

Programmatic access to onward-ticket bookings via MCP and REST. Used by AI agents (ChatGPT, Claude, Cursor) and human developers.

Quickstart

API Reference

Interactive OpenAPI 3.1 spec — every public REST endpoint, request/response shape, and try-it-now form.

MCP Tool Reference

Auto-rendered from the live server card. Inputs, examples, and JSON-RPC request/response bodies for each tool.

Recipes

Worked examples: quoting, placing paid orders, looking up status, and grounding LLMs with the help center.

Try it Live

Interactive runner — pick a tool, edit the JSON arguments, see the live response. No keys required.

Agent Discovery

Web Bot Auth, x402, ACP, UCP, MPP — every agent-readiness signal with well-known URLs, status, and Claude Desktop config blocks.

Quick connect

OnwardTicket.us speaks the Model Context Protocol over SSE. Add it to any MCP-compatible client.

Claude Desktop

Modern (Streamable HTTP — preferred):

{
  "mcpServers": {
    "onwardticket": {
      "url": "https://onwardticket.us/api/mcp"
    }
  }
}

Clients that speak only stdio — via mcp-remote proxy:

{
  "mcpServers": {
    "onwardticket": {
      "command": "npx",
      "args": ["mcp-remote", "https://onwardticket.us/api/mcp"]
    }
  }
}

Restart Claude Desktop. The six OnwardTicket tools become available in any chat.

ChatGPT (Pro/Team/Enterprise)

ChatGPT supports MCP servers as Custom Connectors. In Settings → Connectors → Add custom connector, paste:

https://onwardticket.us/api/mcp/sse

Authentication: none. Tools appear automatically once the connector is enabled in your active conversation.

Custom Connector support requires an active ChatGPT Pro, Team, or Enterprise plan.

Cursor / VS Code

Add to your MCP-aware IDE config (e.g. ~/.cursor/mcp.json or Continue's config.json):

{
  "mcpServers": {
    "onwardticket": {
      "url": "https://onwardticket.us/api/mcp"
    }
  }
}

Reload the IDE. The OnwardTicket tools surface alongside your other configured MCP servers.

Rate Limits & API Keys

All public endpoints are rate-limited by IP: 60 requests / minute for general endpoints; 10 requests / minute for sensitive endpoints (login, password reset). /.well-known/* discovery endpoints are exempt — AI crawlers may poll them freely.

Developer-facing REST endpoints (/api/services, /api/quote, /api/orders/lookup) support optional API key gating. In development (no API_KEYS env var set) the guard is a no-op. In production, pass keys via X-API-Key header or ?api_key= query param. See Agent Discovery → API Keys for details.

Reference links

/docs/agents — agent discovery signals (Web Bot Auth, x402, ACP, UCP, MPP)
/agents — high-level integration overview
/.well-known/mcp/server-card.json — live MCP server descriptor (SEP-2127)
/api/openapi.json — OpenAPI 3.1 spec
/llms.txt — curated index for AI crawlers
/llms-full.txt — full content corpus for AI ingestion

Need help integrating? Email [email protected]with the subject line “agent integration”.