LLM Chain with Structured Output Parser and Auto-Fixing

How it works

1. 1.Manual Execute Workflow trigger starts the run.
2. 2.A Set node stores the fixed question in the input field.
3. 3.The prompt flows to an LLM Chain powered by OpenAI Chat Model at temperature 0.
4. 4.A Structured Output Parser defines a JSON schema of state plus an array of cities with numeric population.
5. 5.An Auto-fixing Output Parser wraps that schema and uses a second OpenAI model to repair non-conforming outputs.

Use cases

• Generating schema-valid geographic reference data for downstream processing.
• Demonstrating reliable structured LLM output with automatic validation and repair.

Requirements

• An n8n instance with the LangChain nodes available.
• Valid OpenAI API credentials attached to both Chat Model nodes.
• No user configuration needed; the prompt text is hardcoded in the Set node.

Setup and testing in n8n

1. Execute the workflow manually from the n8n editor.
2. Verify the result matches the defined JSON schema.
3. Temporarily break the schema to observe the auto-fixing parser behavior.

Service Value

Ideal as a smart automation service combining integrations and AI to produce ready-to-use results.