Universal JSON Fix

Repairs broken or malformed JSON strings, transforming them into valid JSON. It's particularly useful for handling inconsistent JSON output from Large Language Models (LLMs) and other error-prone sources.

Overview

This actor fixes common JSON errors like missing quotes, brackets, and commas; mixed single/double quotes; improperly escaped characters; trailing commas; and JavaScript-style comments. It also cleans JSON wrapped in markdown code blocks or surrounded by extra text, which is typical in LLM responses.

Key Features

• LLM Specialist: Handles malformed JSON from ChatGPT, Claude, Llama, and other AI models.
• Broad Error Correction: Fixes structural errors (missing/broken syntax) and formatting issues (mixed quotes, comments).
• Non-Destructive: Aims to preserve the original data structure and values.
• Self-Contained: Zero external dependencies.
• Deterministic: Provides consistent repair results.

How to Use

Provide the malformed JSON string as input. The actor processes it and returns valid, formatted JSON. No configuration is needed.

Input/Output

Input: A text string containing malformed JSON.
Output: A valid JSON object or array, repaired and standardized.

Example Scenario:
You have this LLM output:

{ name: 'John', age: 30, cities: ['NYC', 'London',] }

The actor will return:

{"name": "John", "age": 30, "cities": ["NYC", "London"]}

Use Cases

• Parsing JSON from AI/LLM APIs.
• Cleaning JSON in data pipelines before processing.
• Handling inconsistent JSON from external APIs.
• Debugging and fixing JSON syntax during development.
• Repairing corrupt JSON files during data migration.

Technical Approach

The actor uses a parsing engine that reconstructs JSON using formal grammar rules. It intelligently adds missing structural elements, properly escapes string values, normalizes quote styles, and removes invalid syntax.