Robust JSON Parser for AI Outputs & Dirty Data

This n8n sub-workflow is a powerful tool for handling 'dirty' JSON strings commonly produced by AI language models or extracted from web pages. It systematically cleans input by trimming whitespace and removing Markdown fences, then applies targeted fixes: escaping control characters, correcting backslash sequences, removing trailing commas, and repairing unescaped quotes within strings. If direct parsing fails, it intelligently extracts JSON-like blocks from larger text and reprocesses them.

The benefits are significant for automation workflows: it saves hours of manual data correction, reduces errors in downstream processes, and ensures consistent JSON output even from imperfect sources. By outputting clean, structured data (with optional parsing status for debugging), it integrates seamlessly as a sub-workflow, boosting reliability in AI-driven pipelines.

Ideal use cases include processing AI-generated content for website building (e.g., landing pages), API responses, or scraped data. Use it after AI nodes like OpenAI to structure outputs for databases, CRMs, or further automation. No setup required beyond linking via Execute Sub-Workflow node with a 'text' input.

Setup is quick: add Execute Sub-Workflow in your main flow, select this workflow, and pass { "text": "your JSON string" }. Deactivate the final Set node to retain debug info.