Ask Mario's CV
LiveMario's public CV, selected systems, RubyGems packages, engineering blog, and Reykjavík availability — compiled at build time from this site.
Mario Ruiz Sánchez · Model Context Protocol
The Model Context Protocol (MCP) lets AI clients call structured, read-only tools over the network instead of scraping HTML. I build small public MCP servers on marioruiz.net — Streamable HTTP, JSON-RPC 2.0, no authentication — so Claude, ChatGPT, Cursor, and VS Code can query verified data with citations instead of invented claims.
Each card links to setup details, tool reference, and curl examples for that endpoint.
Mario's public CV, selected systems, RubyGems packages, engineering blog, and Reykjavík availability — compiled at build time from this site.
Live Icelandic earthquakes, volcano VONA/VALS status, weather observations, CAP warnings (IMO), and road conditions (Vegagerðin) with fetched_at_utc and stale flags.
More servers coming.
Examples below use the Ask Mario's CV endpoint (https://marioruiz.net/mcp/cv). For Iceland live, replace the URL with https://marioruiz.net/mcp/iceland and adjust the server name.
Add to .cursor/mcp.json in your project (or global Cursor MCP settings):
{
"mcpServers": {
"marioruiz-cv": {
"url": "https://marioruiz.net/mcp/cv"
}
}
}
See Cursor MCP.
claude mcp add marioruiz-cv --transport http https://marioruiz.net/mcp/cv
.mcp.json{
"mcpServers": {
"marioruiz-cv": {
"type": "http",
"url": "https://marioruiz.net/mcp/cv"
}
}
}
Claude Desktop can add a remote connector with the same HTTPS URL (Settings → Connectors). See Claude Code MCP.
/mcp in the composer to confirm the server is connected.On ChatGPT web, remote MCP tools are installed through plugins or workspace connectors — the desktop app and Codex read the same ~/.codex/config.toml entries. See OpenAI MCP docs.
https://marioruiz.net/mcp/cv
Add to .vscode/mcp.json (root key "servers", not mcpServers). Use Agent mode in Copilot chat:
{
"servers": {
"marioruiz-cv": {
"type": "http",
"url": "https://marioruiz.net/mcp/cv"
}
}
}
See VS Code MCP servers.
Codex stores MCP servers in ~/.codex/config.toml (shared with the ChatGPT desktop app and Codex IDE extension):
[mcp_servers.marioruiz-cv] url = "https://marioruiz.net/mcp/cv"
Or via CLI:
codex mcp add marioruiz-cv --transport http https://marioruiz.net/mcp/cv
See OpenAI Codex MCP.
See Grok Bot — computer and apps.
https://marioruiz.net/mcp/cv
Add an mcpServers entry to ~/.gemini/settings.json or project .gemini/settings.json:
{
"mcpServers": {
"marioruiz-cv": {
"httpUrl": "https://marioruiz.net/mcp/cv"
}
}
}
Use httpUrl for Streamable HTTP endpoints. See Gemini CLI MCP.
Edit ~/.gemini/config/mcp_config.json or workspace .agents/mcp_config.json. Remote servers use Streamable HTTP via serverUrl:
{
"mcpServers": {
"marioruiz-cv": {
"serverUrl": "https://marioruiz.net/mcp/cv"
}
}
}
In Cascade, open Open MCP config file and edit mcp_config.json (~/.config/devin/mcp_config.json on macOS/Linux). Remote HTTP servers need serverUrl or url:
{
"mcpServers": {
"marioruiz-cv": {
"serverUrl": "https://marioruiz.net/mcp/cv"
}
}
}
See Windsurf Cascade MCP.
Test the live endpoint without a full client (requires Node.js 22.23.4+, 24.22.1+, or 26.12.1+):
npx @modelcontextprotocol/inspector --server-url https://marioruiz.net/mcp/cv --transport http
The web UI opens in your browser; use tools/list and tools/call to probe the server. See MCP Inspector.