n8n-io/n8n

Why It's Trending

n8n has crossed 204K stars and is trending because of a significant architectural expansion: native AI capabilities including MCP client/server support. [1] This positions n8n not just as a REST API connector (its original value proposition) but as a self-hosted AI agent orchestration platform.

The timing aligns with a broader market shift: teams building AI-augmented workflows are choosing between cloud automation platforms (Zapier, Make) and self-hosted options (n8n). n8n's open-source, self-hostable model combined with LLM-native nodes has made it the default choice for privacy-conscious teams that need both workflow automation and AI integration.

Technical Overview

n8n's TypeScript architecture is a node-based visual workflow builder where each node encapsulates an integration or transformation. [1] The new AI nodes treat LLM API calls as first-class workflow steps — chain a trigger, an LLM call, a conditional, and a database write without leaving the workflow canvas.

The MCP integration is the most technically interesting addition: n8n can expose workflows as MCP servers (any Claude Code session can invoke them as tools) and can consume MCP servers as n8n node types. This bidirectional integration means n8n workflows and Claude Code agents can directly call each other, creating a hybrid automation/agent architecture that neither tool could achieve alone.

For Understory Labs

bud's current architecture routes Gmail → n8n → FastAPI classifier → Supabase. The n8n MCP integration opens a direct question: can the FastAPI classification step be replaced with an n8n Claude MCP tool call?

If n8n can invoke Claude via MCP natively, the Python FastAPI service becomes optional for classification tasks — n8n calls Claude with the email content, Claude returns a classification JSON, n8n routes to Supabase. This would reduce the bud infrastructure from three services (n8n + FastAPI + Supabase) to two (n8n + Supabase) for the core ingestion pipeline. Worth evaluating in the next bud architecture review.

Direct operational dependency for bud with meaningful architectural implications — the MCP integration could simplify the pipeline by eliminating the FastAPI classification layer. Review the MCP node documentation before the next bud architecture session.