Power your agent
The publisher's own URL. Not a PlugMCP endpoint.
Host a copy with our keys and logs.
https://mcp.theholisticcare.com/mcpThe Claude button above opens Claude with the name and URL filled in. You still confirm it there.
claude mcp add --transport http com.theholisticcare-open-mindfulness https://mcp.theholisticcare.com/mcpRun in a terminal.
{ "mcpServers": { "com.theholisticcare-open-mindfulness": { "command": "npx", "args": [ "-y", "mcp-remote", "https://mcp.theholisticcare.com/mcp" ] } }}claude_desktop_config.json
Remote URL or a command on your machine? How the two installs differ
Run on PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
LaunchFrom the publisher
A public, read-only Model Context Protocol server for The Holistic Care's approved public mindfulness resources — interactive games, free guided audio practices, plain-language research summaries, a glossary, and the public PanchaVikas framework overview.
Read-only mindfulness games, guided practices, research, glossary and PanchaVikas resources.
https://mcp.theholisticcare.com/mcp (Streamable HTTP)api.theholisticcare.com (docs, source). This server never connects to Sanity CMS, a database, or any payment system directly.The Holistic Care already publishes a public REST API of its mindfulness content. This server is a thin MCP wrapper around that API so any MCP-compatible client (Claude Desktop, Claude Code, custom agents, etc.) can discover and use that content through native tool calls instead of a bespoke HTTP integration — with the same content boundaries the REST API itself already enforces.
MCP client (Claude, etc.)
│ Streamable HTTP, JSON-RPC
▼
thc-mindfulness-mcp (this repo — Next.js App Router, deployed to Vercel)
│ plain HTTPS GET, no credentials
▼
api.theholisticcare.com (sibling REST API project)
│
▼
Sanity CMS (never touched directly by this server)
@modelcontextprotocol/server v2 (createMcpHandler, McpServer), the modern high-level package. @modelcontextprotocol/sdk v1 is kept only as a dev dependency for lower-level type definitions and future in-process test clients.McpServer instance (src/server.ts) — there is no session store, so nothing about one caller's request can leak into another's.src/schemas.ts), re-validated server-side regardless of what the SDK itself already checks.src/lib/apiClient.ts, src/lib/constants.ts). The base URL is a compile-time constant; no tool argument can ever change what host this server talks to.All seven tools are read-only, take no destructive action (readOnlyHint: true, destructiveHint: false in every tool's annotations), and return a small (default 6, max 20) list of normalized resources rather than a bulk dump.
Every list-returning tool shares the output shape { results: Resource[], count: number }; get_mindfulness_resource and lookup_mindfulness_term return { found: boolean, resource: Resource | null }. See each tool's own file under src/tools/ for its exact Zod input schema and a code comment explaining any filter it cannot apply with full fidelity (documented rather than silently ignored — see the note on /v1/search's narrower parameter set below).
A known, documented limitation: the upstream /v1/search endpoint accepts only q/limit/offset, while /v1/resources and the dedicated per-family endpoints accept richer filters (category, audience, skill, age_min/age_max) but no free-text query. Where a tool needs both free-text search AND a filter /v1/search doesn't support server-side, it over-fetches from /v1/search and applies the filter client-side (see src/lib/filters.ts) — this is called out explicitly in the affected tool's description text so an MCP client (and the person using it) knows the filter is real but applied after the fact, not a guarantee from the upstream API itself.
Five static, informational MCP resources — none of them call the upstream API, so they're always available even if api.theholisticcare.com is temporarily down:
There are no MCP prompts in this server, by design — the spec for this project explicitly avoided adding prompts just because the protocol supports them; nothing here needed one.
Any Streamable-HTTP-capable MCP client can connect directly to the production URL. For a claude_desktop_config.json-style remote MCP entry:
{
"mcpServers": {
"thc-open-mindfulness": {
"url": "https://mcp.theholisticcare.com/mcp"
}
}
}
No headers, tokens, or environment variables are required on the client side.
npm install
npm run dev
The MCP endpoint is then available at http://localhost:3000/mcp, and a health check at http://localhost:3000/health.
No secrets are required for local development. The only environment variable this server reads is THC_API_BASE_URL (see .env.example), and it defaults to the production REST API if unset — set it only to point a local instance at a different API (e.g. a locally running copy of thc-open-mindfulness-api).
npm run typecheck # tsc --noEmit
npm run lint # next lint
npm run test # unit tests (mocked fetch, no network)
npm run test:contract # MCP protocol-level tests, in-process
MCP Inspector is the reference tool for interactively exercising an MCP server's tools and resources.
npx @modelcontextprotocol/inspector
Then connect it to either:
http://localhost:3000/mcp (a local npm run dev instance), orhttps://mcp.theholisticcare.com/mcp (production)using the "Streamable HTTP" transport option and no authentication. You should see all 7 tools and all 5 resources listed, and be able to call each one interactively.
This project deploys to Vercel as a standard Next.js App Router app.
THC_API_BASE_URL defaults to https://api.theholisticcare.com).mcp.theholisticcare.com at the Vercel project (Vercel dashboard → Domains), and add the corresponding DNS record at the DNS provider for theholisticcare.com.https://mcp.theholisticcare.com/health returns {"status":"ok",...}https://mcp.theholisticcare.com/mcp responds correctly to an MCP Inspector session (see above)npm run smoke:production passes (scripts/smoke-production.mts — exercises the live endpoint end-to-end)See SECURITY.md for the full write-up. In short: no secrets, no writes, no direct Sanity/database access, SSRF-hardened outbound requests, Host/Origin header validation on the HTTP transport, and a stateless request model. Report a vulnerability to [email protected] rather than opening a public issue.
See LICENSING.md for the full explanation. Short version: the code in this repository is MIT licensed; the mindfulness content it retrieves and returns (games, practices, research summaries, glossary entries, blog excerpts, PanchaVikas overviews) is Copyright © The Holistic Care, all rights reserved, unless a specific resource says otherwise — it is public to read and reference, not to redistribute wholesale. See also ATTRIBUTION.md for how to credit The Holistic Care when you build on top of this server.
This is a small, single-purpose server maintained alongside The Holistic Care's public API and website. Bug reports and pull requests are welcome via GitHub Issues. For anything security-related, see SECURITY.md instead of opening a public issue.
© 2026 MCP Directory.
Read-only mindfulness games, guided practices, research, glossary and PanchaVikas resources.
{ "mcpServers": { "com.theholisticcare-open-mindfulness": { "url": "https://mcp.theholisticcare.com/mcp", "headers": {} } }}Tools
No tool list yet. Remote servers get one after the first health check.
FAQ
Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
No required environment variable is listed on this record.
THC Open Mindfulness MCP is reached over HTTP.
Last checked 2026-11-08. Read-only mindfulness games, guided practices, research, glossary and PanchaVikas resources.