condense-json 1.0

Simon Willison's Weblog reports that condense-json, a utility for compressing repetitive JSON data, has reached version one-point-zero. The tool identifies duplicated strings or substrings and replaces them with compressed markers that can be reversed on demand. It's particularly useful for reducing the file size of logs created by large language model applications, where redundant text can quickly balloon storage. After eighteen months of real-world use, the release marks a stable, production-ready point for this tool already embedded in Willison's own LLM logging infrastructure.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton