n8n AI Agents For solo builders

Build AI workflows that stay useful after the demo.

Claude Code and Codex build. n8n runs. A human approves. Ship Lean shows the system.

This hub is the practical route through n8n AI agents: definitions, workflow examples, approval gates, free tools, and the handoff between coding agents and repeatable automation.

Map your workflow -> See the pattern
ship-lean ~/agent-stack

$ split workflow

builder=claude-code runner=n8n approval=human

$ ship safe-agent

ok one judgment step. no agent soup.

STACK_RULE.MD

Use each tool for its lane

The mistake is asking n8n to be a coding agent or asking Claude Code to be a scheduler. The leverage is the handoff.

trigger judgment approval

Start here

The smallest useful n8n AI agent library.

These are the pages Google is already testing Ship Lean against. The job now is to make the cluster impossible to misunderstand.

DEFINITION_01.MD

What is an n8n AI agent?

A plain-English explainer for the AI Agent node, tools, context, decisions, and approval gates.

Open -> TUTORIAL_02.MD

Build an n8n AI agent

The practical pattern: n8n triggers and routes, the model judges, a human approves risky actions.

Open -> WORKFLOW_03.MD

Human approval workflow

A copyable workflow shape for content, leads, support, Search Console, and operations.

Open -> TOOL_04.MD

Workflow builder

Map the trigger, context, agent job, approval gate, and output before wiring the canvas.

Open ->

The modern stack

n8n is future-proof when it is the workflow layer.

The flashy tool will change. The need for triggers, routing, logs, retries, credentials, and approval gates will not.

01 Claude Code / Codex Build and judgment: Read repo context, draft prompts, create diffs, and improve the system.
02 n8n Trigger and routing: Run schedules, listen for events, gather context, call services, retry, log, and notify.
03 Human Approval: Protect voice, money, customers, publishing, production, and anything with real consequences.

Comparison pages

Pick the right tool before you build the workflow.

These pages turn "which tool should I use?" into a durable decision system for AI builders.

Codex vs n8n

Use Codex for repo-aware building. Use n8n for recurring triggers, credentials, retries, and run history.

Compare ->

AI coding agent vs workflow automation

The broader decision rule for Claude Code, Codex, Cursor, n8n, Zapier, and Make.

Compare ->

n8n AI Agent vs Zapier AI Actions

Pick the right runner when speed, app coverage, self-hosting, agent depth, and control matter.

Compare ->

Decision rule

When to use n8n AI agents

Use an n8n AI agent when a repeatable workflow has one judgment-heavy step: scoring, classifying, summarizing, drafting, routing, or deciding between fuzzy options.

Do not put your whole business inside one giant agent prompt. Keep the agent scoped. Let n8n own the run history, credentials, retries, and notifications. Put a human approval gate before anything public, expensive, customer-facing, or production-facing.

Recommended path

  1. Read what an n8n AI agent is.
  2. Use the workflow builder.
  3. Copy the human approval workflow pattern.
  4. Use the Claude Code + n8n planner if the workflow needs repo-aware work.

FAQ

Is n8n an AI agent platform?

n8n is primarily a workflow automation platform. It becomes useful for AI agents when it gives the model triggers, tools, context, routing, memory, logs, and approval gates.

Should solo builders learn n8n or Claude Code first?

Learn the job split first. Claude Code and Codex are better for building and repo-aware judgment. n8n is better for recurring triggers, integrations, routing, retries, and approvals.

What is the safest first n8n AI agent workflow?

Start with one low-risk judgment step: classify a lead, score a Search Console query, summarize a support request, or draft a content brief. Add human approval before anything public or customer-facing.