Am I doing something wrong with Auto-fixing Output Parser

Describe the problem/error/question

When I have an agent node that needs an structured output, sometimes the LLM will give it to me in a code bubble or the format would not be perfect, for that I thought we had the “Auto-fixing Output Parser” but I never see the model part of it activating, and I think in cases like this it should definetly do


As you can see it moves from having a code bubble with the json to empty answer…

json definition is:

strong text{“type”: “object”,
“properties”: {
“comunication_update”: {
“type”: “string”,
“description”: “Details of the communication update”
},
“id”: {
“type”: “integer”,
“description”: “Unique identifier for the record”
},
“Status”: {
“type”: “string”,
“description”: “Current status of the record”
},
“Stage”: {
“type”: “string”,
“description”: “Current stage of the process”
}
},
“required”: [“comunication_update”, “id”, “Status”, “Stage”]
}

How is it that it doesn’t work? is this a bug?

Please share your workflow

{
  "nodes": [
    {
      "parameters": {
        "options": {}
      },
      "id": "f8855e94-f6c2-47bf-a8c9-965b21c6e110",
      "name": "Auto-fixing Output Parser",
      "type": "@n8n/n8n-nodes-langchain.outputParserAutofixing",
      "typeVersion": 1,
      "position": [
        1960,
        720
      ]
    },
    {
      "parameters": {
        "options": {}
      },
      "id": "bb01fc25-9993-4a01-8e4e-fcf8a5dae554",
      "name": "OpenAI Chat Model1",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "typeVersion": 1,
      "position": [
        1920,
        920
      ],
      "credentials": {
        "openAiApi": {
          "id": "yv064MBJLbwsL817",
          "name": "OpenAi fix output parser"
        }
      }
    },
    {
      "parameters": {
        "schemaType": "manual",
        "inputSchema": "{\"type\": \"object\",\n  \"properties\": {\n    \"comunication_update\": {\n      \"type\": \"string\",\n      \"description\": \"Details of the communication update\"\n    },\n    \"id\": {\n      \"type\": \"integer\",\n      \"description\": \"Unique identifier for the record\"\n    },\n    \"Status\": {\n      \"type\": \"string\",\n      \"description\": \"Current status of the record\"\n    },\n    \"Stage\": {\n      \"type\": \"string\",\n      \"description\": \"Current stage of the process\"\n    }\n  },\n  \"required\": [\"comunication_update\", \"id\", \"Status\", \"Stage\"]\n}"
      },
      "id": "5cb740bb-9bcc-4b9d-934d-b448b98f5f62",
      "name": "Structured Output Parser",
      "type": "@n8n/n8n-nodes-langchain.outputParserStructured",
      "typeVersion": 1.2,
      "position": [
        2080,
        920
      ]
    }
  ],
  "connections": {
    "Auto-fixing Output Parser": {
      "ai_outputParser": [
        []
      ]
    },
    "OpenAI Chat Model1": {
      "ai_languageModel": [
        [
          {
            "node": "Auto-fixing Output Parser",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "Structured Output Parser": {
      "ai_outputParser": [
        [
          {
            "node": "Auto-fixing Output Parser",
            "type": "ai_outputParser",
            "index": 0
          }
        ]
      ]
    }
  },
  "pinData": {},
  "meta": {
    "templateCredsSetupCompleted": true,
    "instanceId": "5aa5d2a1603df28859474f270c5663bf0de5fdf9d6c795d12d01619750c46ce4"
  }
}

## Share the output returned by the last node
<!-- If you need help with data transformations, please also share your expected output. -->

## Information on your n8n setup
- **n8n version:** 1.76.3
- **Database (default: SQLite):**  Postgres
- **n8n EXECUTIONS_PROCESS setting (default: own, main):**queue
- **Running n8n via (Docker, npm, n8n cloud, desktop app):** online deployed through Docker
- **Operating system:**: windows

same problem here, this node doesn’t fix anything, bad output is propagated further

1 Like

Also having the exact same issue with this node, can anyone help with this one?

1 Like

I’m having this exact issue. This makes my ‘respond to webhook’ node return an empty ‘outpt’ object.

1 Like

Same issue also. I thought maybe it was my configuration, but have spent hours trying to get the Auto-Fixing Output parser to rerun after it gets an error and it doesn’t ever rerun.

1 Like

Same here, bumping this topic.

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.