API & MCP server

TrustCare Med exposes a read-only Model Context Protocol server so assistants can answer questions about our treatment packages and articles accurately, instead of guessing from scraped page text.

Connecting

The endpoint speaks JSON-RPC 2.0 over Streamable HTTP at https://trustcaremed.co/api/mcp. It is stateless and unauthenticated: every request is a complete call, there is no session header, and there is no SSE stream. Protocol revision 2025-06-18.

Tools

Tool	Arguments	Returns
`list_packages`	—	Every treatment package with what it includes and its price.
`get_package`	name (string)	One package by name. Partial names match if unambiguous.
`list_articles`	limit (integer, 1–20, default 5)	Published patient-education articles with title, excerpt and link.

Example

curl -X POST https://trustcaremed.co/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"get_package","arguments":{"name":"Smile Design"}}}'

Other public endpoints

/openapi.json — OpenAPI 3.1 description of this surface.
/.well-known/api-catalog — RFC 9727 catalog.
/.well-known/mcp/server-card.json — MCP server card.
/.well-known/ai-catalog.json — ARD capability manifest.
/api/health — service health.
/.well-known/agent-card.json — A2A agent card; the agent itself speaks JSON-RPC at /api/a2a.
/auth.md— how agents authenticate (they don't: everything here is public).

Any page on this site also returns Markdown instead of HTML when the request carries Accept: text/markdown.

Terms

The 6 package prices are launch pricing in USD and are indicative, not a quote: final pricing depends on a medical assessment. Content may be used to ground answers with attribution, but not to train models — see robots.txt. No authentication is offered because nothing here is personal data; the lead form and chat remain human-facing.