What is the Comprehensive Extraction Principle?

It is a mandate that the skill must err toward extraction rather than rejection. For any domain-relevant source, the skill aims to skip less than 10% of the content to ensure all evidence and reasoning paths are captured.

Does it prevent duplicate notes from being created?

Yes, it uses vector search and grep checks to identify existing notes. If a similar claim is found, it creates an 'enrichment' task rather than a duplicate note.

How does the skill handle very large documents?

The skill includes a 'Source size check' that automatically triggers chunked processing for files over 2500 lines, ensuring no insights are lost due to context window limits.

Can I customize the types of insights extracted?

Yes. By configuring the 'ops/derivation-manifest.md' file, you can define custom vocabulary, extraction categories, and domain-specific processing rules.

How does 'reduce' differ from standard AI summarization?

While summarization condenses text, 'reduce' extracts specific, atomic reasoning and claims that can stand alone as distinct, retrievable notes for a knowledge graph.

Knowledge Reduction & Insight Extraction

Name: Knowledge Reduction & Insight Extraction
Author: LopeWale

byLopeWale

0•

Learning & Documentation

Extracts structured, atomic insights and articulated reasoning from raw source material with a high-comprehension default.

The Reduce skill is a sophisticated extraction engine designed to transform raw text, documentation, and source files into a structured graph of atomic propositions. Unlike standard summarization, it focuses on 'articulated reasoning'—the underlying 'why' behind implementations—ensuring that even subtle evidence, tensions, and patterns are captured as retrievable notes. It operates on a Comprehensive Extraction Principle where skipping more than 10% of domain-relevant material is considered a bug, making it the ideal tool for building deep, searchable knowledge bases, second brains, and technical manifests directly within a development workflow.

Key Features

010 GitHub stars

02Configurable domain-specific vocabulary mapping via derivation manifests.

03Automatic chunking and processing for large source files exceeding 2500 lines.

04Automated classification of insights into Tensions, Patterns, Anti-patterns, and Enrichments.

05Intelligent duplicate detection using vector search and keyword analysis.

06Comprehensive extraction with a <10% skip rate for domain-relevant material.

Use Cases

01Building and maintaining a project-specific 'Second Brain' or internal knowledge graph.

02Mining research papers or meeting transcripts for trade-offs and underlying reasoning.

03Converting dense technical documentation into atomic, searchable project notes.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lopewale/ampllabs reduce

For use in Claude.ai and ChatGPT