Distill Semantic Compression
Reduces noise and redundancy in markdown and text files while preserving all factual information and code.
The Distill skill is an information-theoretic text editor designed to maximize facts-per-word by removing structural padding, hedging, and buzzwords. It utilizes a multi-step process combining deterministic regex patterns and semantic LLM refinement to clean up documentation, notes, and prose. This skill is particularly useful for improving the signal-to-noise ratio in AI-generated content or bloated technical documentation, ensuring that every word earns its place without losing essential data, code snippets, or URLs.
主要功能
01Deterministic pre-processing to eliminate common verbose patterns and hedges
02Built-in verification gates to prevent information loss
03Semantic compression that preserves all facts, numbers, and code blocks
04438 GitHub stars
05Recursive directory processing for bulk documentation cleanup
06Automated measurement of word count and gzip compression improvements
使用场景
01Improving the readability and density of internal project notes and READMEs
02Reducing the size of large text datasets for better context window management
03Cleaning up verbose or repetitive AI-generated technical documentation