In this final project, your Slack agent hands off all decision-making to n8n. It sends every question to a webhook, and n8n handles the orchestration: calling the RAG system first, then escalating if needed. You’ll build a multi-step fallback chain that mirrors how real AI agents are wired in production — including creating Jira tickets when answers fail.

You will:

• Forward Slack messages from your agent to an n8n webhook
• Build an n8n workflow that calls your MCP server’s /rag endpoint
• Evaluate confidence or quality of RAG answers inside the workflow
• Add a fallback: call another MCP endpoint (like /tools or /postgres) when RAG fails
• Escalate unanswered questions to Jira Cloud by creating real support tickets
• Post the final response back into the Slack thread — whether it’s an answer or a fallback notice