Prompt Chain

Automate Data Flows with N8N

Name: Automate Data Flows with N8N
Availability: OnlineOnly
Author: Promptfoo

N8N Agent is a multi-step prompt workflow example demonstrating agent-based automation patterns for the n8n workflow platform, ready to run and test.

Copy chain

Works with n8n

Promptfoo

Maintainer?

Spark score

out of 100

Updated yesterday

Version code-scan-action-0.1

Models

universal

Add to Favorites

Why it matters

Orchestrate and automate complex data workflows by integrating with N8N. This asset enables seamless data extraction, transformation, and loading across various applications and databases.

Outcomes

What it gets done

Connect to N8N to trigger automated workflows.

Extract data from various sources.

Perform ETL (Extract, Transform, Load) operations.

Query and move data between systems.

Install

Add it to your toolbox

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/pfoo-n8n-agent | bash

Capabilities

What this chain does

Extract

Pulls structured data fields from unstructured text.

ETL & sync

Moves and transforms data between systems on a schedule.

Query a database

Writes and executes SQL or NoSQL queries on databases.

Overview

N8N Agent

What it does

N8N Agent is a runnable example of a multi-step prompt workflow (prompt_chain) designed for the n8n automation platform. It demonstrates how to structure agent-based prompt orchestration within n8n, providing a concrete reference implementation that can be executed immediately to see prompt chains in action.

How it connects

Use this example when you're building agent workflows on n8n and need a working reference for multi-step prompt patterns. It's ideal for developers prototyping automation sequences or learning how to integrate prompt chains into n8n's workflow engine.

Source README

n8n-agent (n8n AI Agent Evaluation)

You can run this example with:

npx promptfoo@latest init --example n8n-agent

This example demonstrates how to evaluate n8n AI agents and workflows using the n8n provider.

Prerequisites

A self-hosted n8n instance with a webhook-triggered workflow
An AI agent workflow that accepts messages and returns responses

Setup

Create an n8n workflow with a Webhook trigger node
Add your AI agent logic (e.g., AI Agent node, OpenAI node)
Configure the workflow to return the agent's response
Update promptfooconfig.yaml with your webhook URL:

providers:
  - id: n8n:https://your-n8n-instance.com/webhook/your-agent-id

Environment Variables

export N8N_API_KEY=your-api-key  # If your webhook requires authentication

Running the Example

# Run the evaluation
npx promptfoo eval

# View results
npx promptfoo view

Configuration Options

url: Webhook URL (alternative to specifying in provider path)
method: HTTP method (default: POST); GET sends rendered body fields as query parameters
headers: Additional request headers
body: Custom request body template with Nunjucks support; prefer object form for JSON payloads
transformResponse: JavaScript expression to extract output
sessionHeader: Request header name for the session ID
sessionParser: JavaScript expression to extract a session ID
sessionField: Request body field for a supplied session ID

For multi-turn runs, pass sessionId in test variables or use a multi-turn strategy so sessions
remain scoped to one conversation. Keep webhook credentials in environment-backed headers rather
than embedding them in webhook URLs. The provider uses a fingerprinted display ID and skips
response caching so tokenized URLs and session-bearing requests do not enter response-cache
diagnostics or storage; URLs still remain in configuration and outbound requests.

Response Formats

The provider automatically handles common n8n response formats:

{ "output": "Response text" }
{ "response": "Agent response" }
{ "message": { "content": "Hello" } }
[{ "json": { "output": "Result" } }]

For custom formats, use transformResponse:

config:
  transformResponse: 'json.data.agentMessage'

For more information, see the n8n Provider documentation.

Discussion