The pattern that works: build the capability once as a FastAPI backend, then use FastAPI-MCP (from the Tadata team) to expose it as an MCP server in a few lines of code - both the REST API and the MCP tools live in a single deployment. From there, n8n's MCP Client node connects to it and your workflow can call those tools directly.
The real payoff is reusability, not novelty: the same FastAPI-MCP servers (Docker-containerized) get consumed by n8n, AI agents, Flowise, a form-based Flask UI, and Excel via xlwings Lite - no per-client rewrite. A worked example: pull price data from Yahoo Finance, compute technical indicators, feed the chart into Gemini Vision for AI chart analysis, and emit a PDF plus a web report - all orchestrated in n8n over MCP.
One honest caveat: MCP has real vulnerabilities and deserves battle-testing before you expose anything sensitive - treat a public MCP server as an attack surface, not a convenience.
The servers are public - connect or clone. Catalog + setup: https://www.tigzig.com/apis (the technical-analysis server: https://www.tigzig.com/apis/technical-analysis; unified data MCP endpoint: https://api.tigzig.com/mcp). Post: https://www.tigzig.com/post/build-ai-workflows-mcp-servers-n8n-technical-analysis. Related: how to build an MCP server https://www.tigzig.com/agents-faq/how-to-build-an-mcp-server and securing one https://www.tigzig.com/agents-faq/how-to-secure-a-database-for-ai-agents-and-mcp.