About this eBook

n8n is the rare automation platform that hands you the whole engine. You can self-host it, read the execution log line by line, drop into a Code node when the visual builder runs out of road, and keep customer data inside your own network.

This book treats n8n as what it has become — an agent runtime. You will build agents that call tools, remember conversations, query a vector store, hand work to sub-agents, and recover from the API that returns a 500 at three in the morning.

Included: eleven importable workflow JSON files, a hardened docker-compose stack with Postgres and queue mode, and a credential-handling chapter written for someone who has to pass a security review.

What you will be able to do

• Self-host n8n on Docker with Postgres, queue mode and sane backups
• Master the AI Agent node — tools, memory, output parsers and system messages
• Turn any workflow into a callable tool the agent can invoke with arguments
• Build retrieval over your own documents with a vector store and proper chunking
• Design multi-agent flows where a router hands work to specialist sub-agents
• Handle failures properly: retries, error workflows, dead letters and idempotency
• Keep credentials, PII and prompts inside infrastructure you control

Table of contents

1. Why self-hosting matters — data residency, cost curves and vendor risk
2. Installing n8n: Docker Compose, Postgres, queue mode, reverse proxy and TLS
3. Workflow fundamentals for agent builders — items, expressions and the Code node
4. The AI Agent node dissected: models, system message, and the tool-calling loop
5. Tools that do real work — HTTP Request tool, Workflow tool, Code tool
6. Memory: buffer windows, Postgres chat memory, and when memory hurts you
7. Vector stores and retrieval — chunking, embeddings, hybrid search, re-ranking
8. Triggers: webhooks, chat, schedules, email and message queues
9. Multi-agent architecture — routers, specialists and the hand-off contract
10. Error handling, retries, idempotency keys and dead-letter workflows
11. Testing and versioning: pinned data, environments, and Git-backed workflows
12. Security: credential scoping, secrets, egress rules and audit logging
13. Scaling: workers, concurrency, execution pruning and cost per run

Who it is for

• Technical operators who want automation without a per-task SaaS bill
• Teams with data-residency or compliance constraints that rule out hosted tools
• Developers who like a visual builder but refuse to be trapped inside one

Questions