Adding Structured Output Parser throws an "Unexpected Token" error

[acquia-ankith]

Bug Description

I have created an AI Agent with

• LLM Model: sonnet 4.5
• Tools: 2-3 MCP Servers
• Memory: Simple memory or PostgresDB
• Parser: Structured output parser.

When I execute the AI Agent the node completes the task that is provided in the user prompt but fails after completing with error stating "Unexpected Token 'I', "I'll help" ... is not a valid JSON .

To Reproduce

1. Create a simple workflow with manual trigger, AI Agent node, and any other type of nodes following the AI Agent node.
2. Connect LLM Model, Memory, Tools and Parser to the AI agent node. The exact config I used:

• LLM Model: sonnet 4.5
• Tools: 2-3 MCP Servers
• Memory: Simple memory or PostgresDB
• Parser: Structured output parser.

3. Execute the workflow.
4. Once the AI Agent completes the task provided in the user prompt then we observe an error on as shown below

FYI: I observe the first response from LLM starts as "I'll help you with .... "

Expected behavior

The AI agent node should execute and provide the structured output without causing any error.

Debug Info

Debug info

core

• n8nVersion: 2.8.3
• platform: docker (self-hosted)
• nodeJsVersion: 22.22.0
• nodeEnv: production
• database: sqlite
• executionMode: regular
• concurrency: -1
• license: enterprise (production)
• consumerId: 6e39d534-70f6-4ad4-89cf-9088b694bc20

storage

• success: all
• error: all
• progress: false
• manual: true
• binaryMode: filesystem

pruning

• enabled: true
• maxAge: 336 hours
• maxCount: 10000 executions

client

• userAgent: mozilla/5.0 (macintosh; intel mac os x 10_15_7) applewebkit/537.36 (khtml, like gecko) chrome/145.0.0.0 safari/537.36
• isTouchDevice: false

Generated at: 2026-02-19T13:52:41.474Z

Operating System

n8n official Docker image "stable" tag

n8n Version

2.8.3

Node.js Version

22.22.0

Database

PostgreSQL

Execution mode

main (default)

Hosting

self hosted

[n8n-assistant]

Hey @acquia-ankith,

Thank you for reaching out! We’ve received your issue and are looking into it. To help us track this internally, we’ve created a Linear ticket with the reference: "GHC-6930".

We’ll keep you updated as we make progress, but if you have any additional details or context to share, feel free to add them here - it’s always helpful!

Thanks for your patience and for bringing this to our attention.

---

For INTERNAL use only: GHC-6930

[tamy-commits]

Hi @acquia-ankith , thank you for the report and the screenshots.

This appears related to how the SOP (Structured Output Parser) processes the model response rather than a failure in the workflow execution itself. According to the documentation, the SOP expects the model to return structured JSON that matches the configured schema. If the model returns natural language instead, the parser attempts to parse the response as JSON and will throw a JSON.parse error.

From the description, the model response seems to begin with text like “I'll help you…”, which would not be valid JSON. To help investigate further, could you please share:

• the schema configured in the Structured Output Parser
• the raw LLM response from the execution data (before parsing)
• the prompt or instructions used for the agent

This will help determine whether the issue originates from the model output format or from how the parser is configured.

Docs:
https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.outputparserstructured/

#triage:needs-info

[tamy-commits]

Thanks for reporting this issue! We’ve attempted to reproduce the problem, but unfortunately, we weren’t able to replicate it. It’s possible there was some environmental or other variable that prevented us from seeing the same behaviour.

If you’re still encountering this issue, please provide any additional details or steps to reproduce it, and we’d be happy to investigate further. Thanks for your understanding!

#closed:cant-reproduce

[n8n-assistant]

Hi @acquia 👋

Thanks so much for your patience while we’ve been improving n8n. We know it’s taken some time to review this issue, and we’re currently updating our triage process to better keep up with our growth.

There have been a lot of improvements in recent releases, so we’d really appreciate it if you could check whether this issue still occurs in the latest release.

When you have a moment, please copy and paste the checklist below as a comment so our system can update the status automatically:

[ ] I have updated to the latest stable version of n8n [please specify version number] and I am still experiencing this issue
[ ] Please close this issue – I am no longer experiencing it
[ ] Please close this issue – [please specify a reason]

If we don’t hear from you in the next 14 days this issue will be automatically closed, feel free to reopen - don’t forget to add the checklist above so that our automated systems notify us of the issue.

Thanks again for helping us keep n8n healthy and up-to-date 💚