Optimize n8n workflow JSON using Azure OpenAI GPT-4o-mini

Reference Document: AI JSON Optimiser Workflow

1. Workflow Overview

The AI JSON Optimiser is a specialized automation designed to refine and improve n8n workflow JSON files. It accepts an existing workflow via a POST webhook, processes it through an AI agent powered by Azure OpenAI (GPT-4o-mini), and returns a cleaned, optimized, and import-ready version of the workflow as a downloadable file.

The workflow is organized into three main functional blocks:

1.1 Input & Extraction: Receives the raw JSON data and validates the payload structure.
1.2 AI Optimization Engine: Uses a Large Language Model (LLM) to perform structural and performance-based improvements on the JSON.
1.3 Output Processing & Delivery: Sanitizes the AI's response, converts it back into a valid n8n file format, and delivers it to the requester.

2. Block-by-Block Analysis

2.1 Input & Extraction

Overview: This block acts as the interface for the workflow. It captures the incoming HTTP request and ensures that the data provided is a valid n8n workflow object before passing it to the AI.

Nodes Involved:
- Receive Workflow JSON (Webhook)
- Extract & Validate Workflow Body (Code)
Node Details:
- Receive Workflow JSON: A Webhook node configured for POST requests. It expects a JSON body. If the request is successful, it waits for the Respond to Webhook node to finish before sending the response.
- Extract & Validate Workflow Body: A JavaScript Code node that performs a safety check. It verifies that body.workflow exists. If the input is empty or malformed, it throws a specific error to prevent the AI from processing null data.

2.2 AI Optimization Engine

Overview: The core of the workflow. This block leverages generative AI to analyze node connections, identify redundancies, and apply best practices for n8n development.

Nodes Involved:
- Optimize Workflow via AI Agent (AI Agent)
- Azure OpenAI Chat Model1 (Azure Chat Model)
Node Details:
- Optimize Workflow via AI Agent: An AI Agent node (LangChain) using a "Define" prompt type. The system message contains strict instructions: return only raw JSON, remove redundant nodes, improve naming, ensure logical positioning (200px spacing), and preserve functionality.
- Azure OpenAI Chat Model1: Connects to the gpt-4o-mini deployment. It provides the intelligence for the optimization tasks.
- Edge Cases: AI might occasionally include markdown code blocks (json ... ) despite instructions, which requires the next block to handle cleanup.

2.3 Output Processing & Delivery

Overview: This block ensures the AI's text output is transformed back into a functional system file. It handles parsing and the final binary file generation.

Nodes Involved:
- Parse, Strip & Validate Optimized JSON (Code)
- Convert Optimized Workflow to File (Convert to File)
- Return Optimized File to Caller (Respond to Webhook)
Node Details:
- Parse, Strip & Validate Optimized JSON: A JavaScript Code node that cleans the AI's string. It trims whitespace, removes markdown backticks if present, and executes JSON.parse(). It also validates that the result contains essential n8n keys like nodes and connections.
- Convert Optimized Workflow to File: Converts the cleaned JSON object into a binary file format (.json).
- Return Optimized File to Caller: Finalizes the HTTP request by sending the binary file back to the user.

3. Summary Table

Node Name	Node Type	Functional Role	Input Node(s)	Output Node(s)	Sticky Note
Receive Workflow JSON	Webhook	Entry Point	(None)	Extract & Validate...	## 📥 Input & Extraction Receives the POST request...
Extract & Validate...	Code	Validation	Receive Workflow JSON	Optimize Workflow...	## 📥 Input & Extraction Receives the POST request...
Optimize Workflow...	AI Agent	AI Processing	Extract & Validate...	Parse, Strip & Validate...	## 🤖 AI Optimization Engine The AI Agent sends the full workflow...
Azure OpenAI Chat Model1	Azure Chat Model	AI Engine	(None)	Optimize Workflow...	⚠️ Azure OpenAI Credentials Required This node requires a valid azureOpenAiApi...
Parse, Strip & Validate...	Code	Data Sanitization	Optimize Workflow...	Convert Optimized...	## 📤 Output Processing & Delivery Strips any accidental markdown...
Convert Optimized...	Convert to File	File Generation	Parse, Strip & Validate...	Return Optimized File...	## 📤 Output Processing & Delivery Strips any accidental markdown...
Return Optimized File...	Respond to Webhook	HTTP Response	Convert Optimized...	(None)	## 📤 Output Processing & Delivery Strips any accidental markdown...

4. Reproducing the Workflow from Scratch

Webhook Setup: Create a Webhook node. Set the method to POST and set Response Mode to When Last Node Finishes.
Input Validation: Add a Code node. Use JavaScript to extract json.body.workflow and add basic if (!workflow) throw Error(...) logic to prevent empty runs.
AI Integration:
- Add an AI Agent node. Set the Prompt Type to Define.
- Inside the Agent, add an Azure Chat Model node. Connect your azureOpenAiApi credentials and set the Model name to gpt-4o-mini.
- In the Agent's System Message, paste instructions requiring the model to return raw n8n JSON, optimize for speed/clarity, and maintain a 200px node spacing.
Cleaning Logic: Add another Code node. Use a script to check if the output is a string, remove ```json tags using .replace(), and then use JSON.parse() to turn it back into an object.
File Conversion: Add a Convert to File node. Set the operation to Convert to JSON (which creates a binary file from the JSON input).
Final Response: Add a Respond to Webhook node. Set the Respond With parameter to Binary File.
Connections: Link the nodes sequentially as described in the summary table.

5. General Notes & Resources

Note Content	Context or Link
Full setup instructions and overview of the internal logic	Included in the "📋 Overview" Sticky Note in the workflow.
Critical: Deployment naming must match `gpt-4o-mini`	Mandatory for the Azure Chat Model node configuration.
Optimal Node Spacing	The AI is instructed to maintain a 200px minimum spacing for visual clarity.

6.8 KiB Raw Blame History