Descriptions:
NetworkChuck demonstrates how to combine n8n workflow automation with Claude Code by connecting them over SSH — a deceptively simple approach that unlocks persistent, session-aware AI agent execution inside any n8n workflow. The core setup requires just three components: an n8n instance, a Linux server running Claude Code (or alternatively Gemini CLI), and SSH credentials configured in n8n’s SSH node. Rather than building a custom HTTP wrapper, the SSH node issues Claude commands directly, with results flowing back into the workflow.
A standout feature covered is session persistence: by generating a UUID in a Code node and passing it as Claude’s `–session-id` flag, users can maintain a continuous conversation across multiple workflow triggers — enabling true back-and-forth interaction rather than one-shot commands. The video then escalates to a more advanced demo where n8n serves purely as the orchestrator, passing a single prompt to Claude Code running in “dangerous mode” with a unified skill file (markdown and Python scripts describing the local network). Claude autonomously spawns three parallel agents to inspect WiFi performance, network throughput, and security posture simultaneously, surfacing a real memory issue on a UDM Pro switch.
Chuck compares Claude Code (requires Pro subscription), Gemini CLI (free with rate limits), and OpenAI Codex for this SSH-driven pattern, and notes that the complexity lives in Claude’s skill files rather than in n8n node sprawl — making previously hour-long workflow builds reducible to a single prompt.
📺 Source: NetworkChuck · Published December 10, 2025
🏷️ Format: Tutorial Demo
