Reduce Knowledge Extractor
Extracts structured insights and atomic reasoning from source material to build comprehensive personal knowledge systems.
Reduce is a specialized extraction engine for Claude Code that transforms raw text into a structured graph of atomic propositions. Unlike standard summarizers that prioritize brevity, Reduce follows a 'Comprehensive Extraction' principle, ensuring that reasoning, evidence, patterns, and tensions are externalized into traversable markdown notes. It is a core component of the arscontexta ecosystem, designed to minimize data loss by maintaining a skip rate of less than 10% for domain-relevant material, effectively turning conversations and documents into a permanent second brain.
Características Principales
01Categorizes insights into distinct types including patterns, anti-patterns, tensions, and implementation ideas.
02Configurable processing depth and vocabulary mapping via local manifest files.
03Comprehensive extraction of core claims, evidence, and validations with minimal skip rates.
042,813 GitHub stars
05Native vector search integration to perform duplicate checks and suggest content enrichments.
06Intelligent large-source handling with automated chunking for documents over 2500 lines.
Casos de Uso
01Mining conversation transcripts to populate a project-specific knowledge vault.
02Converting technical documentation or research papers into atomic, linkable markdown notes.
03Externalizing the underlying 'why' behind architectural decisions in a codebase.