Genten — n8n Setup Guide

⚠️

Before you begin: Make sure Genten is running at localhost:3001. If you're running Genten via Docker, confirm the container is up with docker compose up -d and open http://localhost:3001 in your browser.

Install n8n

You need n8n running on your machine to connect it to Genten. Choose your installation method:

Select your installation method

Not sure which to choose?

Choose npx if:

• You have Node.js 18–24 installed
• You want the simplest setup (1 command)
• You don't use Docker

Choose Docker if:

• You have Node.js 25+ or don't know your version
• You want guaranteed compatibility
• You already use Docker

ℹ️

npx Simpler

Step 1: Check your Node version

node --version

n8n requires Node 18–24. If you see v25 or higher, use Docker instead.

Step 2: Run n8n

npx n8n

▸

Wait for: "Editor is now accessible via: http://localhost:5678"

First run downloads n8n (~1–2 min). Leave this terminal open.

▸

Open http://localhost:5678 and create a local account

💡

Using n8n regularly? Install globally: npm install -g n8n. Then run with n8n start.

ℹ️

Docker Always works

Step 1: Install Docker Desktop

Download from docker.com/products/docker-desktop, install, and open the app. Wait until the menu bar icon shows "Docker Desktop is running."

Docker Desktop menu bar showing Docker Desktop is running

✓ Docker is ready when you see "Docker Desktop is running"

Step 2: Run n8n in Docker

docker run -it --rm --name n8n -p 5678:5678 n8nio/n8n

▸

Wait for: "Editor is now accessible via: http://localhost:5678"

First run downloads the n8n image (~1GB). Leave this terminal open.

▸

Open http://localhost:5678 and create a local account

⚠️

Do NOT use n8n.cloud with Genten. This guide is for n8n running locally only. n8n.cloud cannot reach your local Genten instance without exposing it to the internet, which defeats the purpose of self-hosting.

Register your agent in Genten

In Genten, go to Connect Agent, select n8n, give your agent a name, and click Register My Agent.

Select npx/npm as your installation method to see your connection credentials. Keep this wizard open — you'll use these credentials in Step 4.

Genten showing npx connection credentials

Your connection credentials for npx installation

Select Docker as your installation method. Genten will show host.docker.internal instead of localhost in your credentials — this is required for n8n to reach Genten from inside Docker.

Genten showing Docker connection credentials

Your connection credentials for Docker installation

ℹ️

Why Docker shows different URLs: Docker n8n runs inside a container, so Genten shows host.docker.internal instead of localhost. This ensures n8n can reach Genten from inside the container.

↑ Select an installation method above to see the right credentials

Create a workflow in n8n

Open n8n at http://localhost:5678 and create your first workflow.

▸

Click Create your first workflow

Or click the + button in the top right if you already have workflows

▸

Add a Manual Trigger node

Click the center + → search "Manual Trigger" → select it. This lets you test by clicking a button instead of waiting for a real event.

n8n blank canvas showing where to click to add Manual Trigger node

Click the center + button, then search for "Manual Trigger"

▸

Add an HTTP Request node

Click + to the right of the Manual Trigger node → search "HTTP Request" → select it. This node sends data to Genten.

Search for and select "HTTP Request" to add it to your workflow

Configure the HTTP Request node

Click the HTTP Request node to open its settings.

1. Method: Set to POST

2. URL: Paste from below

http://localhost:3001/api/ingest/n8n

http://host.docker.internal:3001/api/ingest/n8n

↑ Select method above for the correct URL

3. Authentication: Leave as None

4. Send Headers: Toggle ON

5. Header Name: Authorization

6. Header Value: Paste the Bearer acc_... key from your Genten wizard

Add test data

Still in the HTTP Request settings, scroll to Body.

1. Toggle Send Body → ON

2. Set Body Content Type → JSON

3. Click Add Body Field three times and enter:

Name	Value
`event`	`task.completed`
`message`	`Test event from n8n`
`cost`	`0.001`

4. Click Execute Step at the top right, then ✕ to close the panel.

HTTP Request node configuration for Docker

↑ Select method above to see the right screenshot

Verify the connection

After clicking Execute Step in Step 5, check your Genten wizard. The status should change from "Waiting for event..." to Connected.

The connection status shows "Connected" after your first successful test

✓

Connection confirmed. Your n8n workflow is successfully sending data to Genten. Click Finish Setup in the wizard — or continue to Step 7 to enable two-way chat.

Set up two-way chat with AI

Optional

Two-way chat lets you send messages to your n8n workflow from Genten and get AI-powered replies back.

ℹ️

What you'll build: Webhook receives message → AI generates response → webhook replies automatically

Create a new workflow

In n8n, create a new workflow — this is separate from the one you built in Steps 3–5.

Add Webhook node

▸

Click + in the center, search for Webhook, and select it

The Webhook node with test URL ready to copy

Connect test webhook to Genten

▸

In the Webhook node, copy the Test URL

Starts with /webhook-test/

n8n Webhook node showing Test URL and Listen for Test Event button

The Webhook node with Test URL and Listen button ready

▸

In Genten, go to your agent's detail page → Configure Webhooks

Paste the test URL into Test Webhook URL and make sure Test Mode is selected

▸

In n8n, click Listen for Test Event

▸

In Genten, click Test Connection — it should turn green

Webhook node listening for test event with green connection status in Genten

Test connection successful — webhook is listening and Genten confirmed the connection

Configure webhook to respond

▸

In the Webhook node settings, set Respond to When Last Node Finishes

This makes the webhook wait for the AI response before replying to Genten

Webhook node configured with When Last Node Finishes respond setting

Webhook respond setting configured to wait for AI node

Switch to production mode

▸

Copy the Production URL from the Webhook node (starts with /webhook/)

▸

In Genten, paste it into Production Webhook URL and switch to Production Mode

▸

Verify the connection status shows green in Genten

Genten showing production webhook configured with green connection status

Production webhook configured in Genten with green connection status

ℹ️

Running Genten in Docker? If you're running Genten in Docker, you must set the WEBHOOK_BASE_URL and VITE_API_URL environment variables so webhook URLs resolve correctly. See the Docker Setup Guide for details.

Add AI node

▸

Click + to the right of the Webhook node → search Anthropic → select Message a Model

AI node (Anthropic) added and connected to webhook node

Anthropic AI node connected after webhook node

Configure AI node settings

1. Add your Anthropic API key:

▸

Click Credentials → Create New Credential → Anthropic Account

▸

Paste your API key from console.anthropic.com

▸

Click Save

2. Configure the AI model:

Setting	Value
Model	Select your preferred model (e.g. `claude-sonnet-4-6`)
Prompt	`{{ $json.body.text }}`

ℹ️

{{ $json.body.text }} pulls the message text from Genten's webhook payload and sends it to the AI model.

Adding Anthropic API credentials and prompt configuration in n8n

Test and publish the workflow

▸

Back in n8n, click Execute Step on the AI node to test the full workflow

Both nodes should show green checkmarks

Both nodes executed and ready to publish in n8n

Both nodes showing green — ready to publish

▸

Click Publish in n8n to activate the workflow 24/7

Once published, the n8n editor doesn't need to stay open

✓

Production two-way chat is live. Send a message in Genten's Agent Chat — your agent responds 24/7 via n8n + AI.

Successful chat response from n8n agent in Genten

Agent Chat responding via n8n + AI — no editor required

⚠️

Switching back to test mode? You must open your n8n workflow and click Listen for Test Event on the Webhook node, or chat messages will time out.

Troubleshooting

Error	Why it happens	How to fix it
401 Unauthorized	API key wrong or missing	Go back to Step 2, copy the full key, paste as `Bearer YOUR_KEY` in the Authorization header
400 Invalid payload	Request body missing required fields	Make sure your body has both `event` and `message` fields. `cost` is optional.
ECONNREFUSED	Genten isn't running, or wrong URL for Docker	Confirm Genten is running. If using Docker n8n, use `host.docker.internal` not `localhost` in the URL.
503 Service Unavailable	Kill switch is active in Genten	Go to Settings in Genten and turn off the kill switch.
500 Internal Server Error	Genten server crashed or DB is locked	Restart Genten. If error persists, check the Docker logs.
Event not in Activity Feed	Activity Feed only shows live events	Open Activity Feed before executing your workflow, or refresh after.
Changes not saving in n8n	n8n has no visible save button	Use `CMD+S` (Mac) or `CTRL+S` (Windows/Linux).
Wrong URL for installation method	Using localhost with Docker or vice versa	npx → use `localhost`. Docker → use `host.docker.internal`.

FAQ

Test URL — unique to your n8n instance, only active when you click "Listen for Test Event." Changes when you reopen the editor.

Production URL — stable URL, only active when the workflow is published. Use this in Genten once you've confirmed everything works.

n8n outside Docker → Genten outside Docker: use localhost:3001

n8n inside Docker → Genten outside Docker: use host.docker.internal:3001

Both inside the same Docker network: use the service name from docker-compose.yml

Yes. There's no limit on how many tools share the same API key. Usage counts toward the same rate limits and billing. If you want to track n8n usage separately, create a second key at console.anthropic.com.

No. Genten sends messages to your n8n workflow via a webhook endpoint. Telegram is a separate optional channel for OpenClaw agents — it has nothing to do with the n8n integration.

Three nodes: a Webhook that receives messages from Genten, an AI node (Anthropic or OpenAI) that processes the message and generates a response, and the webhook's built-in response that sends the reply back. Without the AI node you can still send a canned response, but the conversation won't be intelligent.

No. Both npx and Docker work. Docker is recommended if your system Node version isn't 18–24, or if you already use Docker for other tools.

Setup Complete Checklist

n8n installed and running at http://localhost:5678

n8n agent registered in Genten, API key saved

HTTP Request node configured with correct URL and Authorization header

Body fields set (event, message, cost)

Workflow executed successfully (green checkmarks in n8n)

Event visible in Genten Activity Feed

(Optional) Two-way chat configured and published

Connect n8n to Genten