Generate AI-Powered Markdown Posts from Workflow JSON with Gemini & LlamaIndex

Description

This AI-powered workflow transforms n8n workflow JSON files into publication-ready, SEO-optimized markdown posts for the n8n community. Simply upload your workflow's JSON, and let Google Gemini 2.5 Pro, guided by a LlamaIndex-powered knowledge base of best practices, automatically generate compelling content.

Why Use This Workflow?

Time Savings: Reduces the time to create a detailed workflow post from over an hour of manual writing to under 2 minutes.

Cost Reduction: Eliminates the need for separate AI content subscriptions or outsourcing content creation tasks.

Error Prevention: Enforces content quality and structural consistency by using a knowledge base of n8n's official guidelines, minimizing formatting errors.

Ideal For

n8n Workflow Creators:** To quickly document and share their creations on the community platform without the tedious, time-consuming writing process.
Developer Advocates:** To standardize and accelerate the production of technical tutorials and workflow showcases.
Content & Marketing Teams:** To streamline the content pipeline for n8n-related blog posts, tutorials, and community engagement initiatives.

How It Works

Trigger: The process starts when you upload an n8n workflow JSON file via a simple web form.
Data Extraction: The workflow automatically extracts the JSON content from the uploaded file.
Intelligence Layer: An advanced AI agent, powered by Google Gemini 2.5 Pro, analyzes the structure, nodes, and metadata of your workflow.
Knowledge Retrieval: The agent consults a specialized, in-memory knowledge base built from n8n's content guidelines. This knowledge base is created by parsing documents with LlamaIndex and refined with a Cohere Reranker for maximum accuracy.
Content Generation: The AI agent synthesizes the technical details from your JSON with the best practices from the knowledge base to write a complete, benefit-driven markdown post.
Output & Delivery: The final, polished markdown content is generated as the workflow's output, ready to be copied and pasted into the n8n community platform.

Setup Guide

Prerequisites

| Requirement | Type | Purpose |
|-------------|------|---------|
| n8n instance | Essential | Workflow execution platform |
| Google Gemini API Key | Essential | Powers the core AI content generation |
| LlamaIndex Cloud API Key | Essential | Parses documents for the knowledge base |
| Cohere API Key | Optional | Improves knowledge base search results |
| Google Drive Account | Optional | For automatically updating the knowledge base from a Google Doc |

Installation Steps

Import the JSON file to your n8n instance.
Configure credentials:
Google Gemini: In the "GEmini 2.5 pro" node, create and add your Google Gemini API credential.
LlamaIndex: In the three HTTP Request nodes named "Parse Document...", "Monitor Document...", and "Retrieve Parsed...", create an HTTP Header Auth credential. The header name is Authorization and the value is Bearer YOUR_LLAMA_INDEX_API_KEY.
Cohere: (Optional) In the "Reranker Cohere" node, create and add your Cohere API credential.
Google Drive: (Optional) If you plan to auto-update the knowledge base, configure Google Drive OAuth2 credentials for the "Knowledge Base Updated Trigger" and "Download Knowledge Document" nodes.
Update environment-specific values:
To use the knowledge base auto-update feature, go to the "Knowledge Base Updated Trigger" node and select the Google Drive file containing your content guidelines.
Customize settings:
The primary system prompt in the "n8ncreator" agent node can be modified to adjust the tone, style, or structure of the generated content.
Test execution:
Run the workflow manually and use the form to upload a sample n8n workflow JSON file to verify that all connections work correctly.

Technical Details

Core Nodes

| Node | Purpose | Key Configuration |
|------|---------|-------------------|
| Form Trigger | Initiates the workflow via a file upload. | Set the "Input Json Workflow" field to required. |
| Langchain Agent | Orchestrates the entire content creation process. | The system prompt contains all instructions for the AI. |
| ChatGoogleGemini | Provides the core generative AI capabilities. | Select your Gemini model of choice (e.g., gemini-2.5-pro). |
| VectorStoreInMemory | Acts as the agent's knowledge base tool. | Configured to use embeddings from a Google Gemini model. |
| HTTPRequest | Interacts with the LlamaIndex API to parse documents. | Set up with LlamaIndex API endpoint and authentication. |

Customization Options

Basic Adjustments:
Change AI Model:** Replace the ChatGoogleGemini node with another LLM node (e.g., OpenAI, Anthropic) to use a different provider.
Adjust System Prompt:** Modify the prompt in the "n8ncreator" node to tailor the output for different platforms (e.g., blog, internal wiki) or change the writing style.

Advanced Enhancements:
Automated Publishing:** Connect the output of the "n8ncreator" node to a Ghost, WordPress, or GitHub node to automatically publish the generated post.
Add Web Search:** Equip the Langchain Agent with a web search tool to allow it to fetch live information about new n8n nodes or services.
Batch Processing:** Replace the Form Trigger with a Read Binary Files node to process an entire folder of workflow JSON files in a single run.

Performance & Optimization

| Metric | Expected Performance | Optimization Tips |
|--------|---------------------|-------------------|
| Execution time | ~1 minute per run | Largely dependent on the Gemini API response time. |
| API calls | 1 LLM call per post | Knowledge base updates trigger LlamaIndex/Google calls separately. |
| Error handling | Built-in retry logic for document parsing | Add an error workflow path after the "n8ncreator" node to handle AI generation failures. |

Troubleshooting

Common Issues:

| Problem | Cause | Solution |
|---------|-------|----------|
| AI output is generic or incomplete | The input JSON file is invalid or lacks key information (e.g., no node names). | Ensure you are uploading a valid, exported n8n workflow JSON. Verify the workflow has been saved with descriptive node names. |
| LlamaIndex parsing fails | The LlamaIndex API key is incorrect or the source document is inaccessible. | Double-check your LlamaIndex API credential. Ensure the Google Doc sharing settings allow access. |
| Credential Error | API keys are missing or incorrect for Gemini, LlamaIndex, or Cohere. | Go to the specified nodes and verify that the correct credentials have been created and selected. |


Created by: khaisa Studio
Category: AI
Tags: AI, Content Generation, Google Gemini, LlamaIndex, Automation
Need custom workflows? Contact us

Connect with the creator:
Portfolio • Workflows • LinkedIn • Medium • Threads