MCP Server
Add AIchitect as a remote MCP server. Get personalised stack recommendations, roast, and challenge your choices — without leaving Cursor or Claude. Because asking your AI to research the entire ecosystem burns tokens. AIchitect already did that work.
https://aichitect.dev/api/mcpQuick setup
Claude Code — add to ~/.claude/settings.json
json{
"mcpServers": {
"aichitect": {
"type": "http",
"url": "https://aichitect.dev/api/mcp"
}
}
}Cursor — Settings → MCP → Add server
Streamable HTTPhttps://aichitect.dev/api/mcpWorks with any MCP-compatible client — Windsurf, Zed, or your own agent.
Tool reference
Try it now
Once the MCP server is added, paste any of these directly into your AI editor.
What AI stack should I use for a RAG app with a small team?Roast my stack: Cursor, Claude Code, LangGraph, SupabaseChallenge my stack: Cursor, LangGraph, SupabaseI'm building an AI coding agent. What stack should I use?