extracting-knowledge-concepts

Extract atomic concepts, relationships, tensions, and uncertainties from documents into structured JSON.

Updated Dec 4, 2025
One-click install
npx skills add https://github.com/dallascrilley/dowser --skill extracting-knowledge-concepts-dallascrilley
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: extracting-knowledge-concepts
Source: https://github.com/dallascrilley/dowser/tree/main/skills/extracting-knowledge-concepts
Command: npx skills add https://github.com/dallascrilley/dowser --skill extracting-knowledge-concepts-dallascrilley

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reading articles, papers, and documents produces unstructured notes that are hard to compare, query, or synthesize later. This Skill converts source text into structured JSON containing atomic concepts, subject-predicate-object relationships, documented tensions between viewpoints, and explicit uncertainties, so knowledge from many sources can be accumulated and analyzed consistently. ## Core Features & Use Cases - Atomic Concept Extraction: Identifies fundamental concepts, techniques, patterns, problems, and tools with canonical names, definitions, confidence levels, and source attribution. - Relationship and Tension Mapping: Captures SPO triples (dependency, hierarchy, conflict, complement, alternative) and preserves contradictions between sources without forcing resolution. - Uncertainty Documentation: Records open questions, why they matter, and what evidence would resolve them. - Use Case: A researcher processing ten papers on distributed systems uses this Skill to produce per-paper JSON files of concepts and relationships, then compares them to find where authors disagree on consistency versus availability tradeoffs. ## Quick Start Use the extracting-knowledge-concepts skill to extract structured concepts, relationships, tensions, and uncertainties from the attached research paper into JSON.

Frequently Asked Questions about extracting-knowledge-concepts

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract structured concepts from research papers?▼

Process each paper through the five-phase methodology: initial scan, concept identification, relationship extraction, tension documentation, and uncertainty documentation. Each concept gets a canonical lowercase name, type, definition, confidence level, and source attribution in JSON format.

What output format does knowledge concept extraction produce?▼

The output is structured JSON with four sections: concepts, relationships (subject-predicate-object triples), tensions, and uncertainties, plus a metadata block with counts and average confidence. A complete template is provided in extraction-output-template.json.

How are contradictions between sources handled during extraction?▼

Contradictions are preserved as tension objects with equal weight given to each position, including supporters and evidence for both sides. The methodology explicitly forbids resolving disagreements by averaging or choosing sides, and suggests experiments that could resolve the tension.

Can I batch process multiple documents for knowledge extraction?▼

Yes, the documentation includes a Node.js example using glob and fs/promises to iterate over markdown files and write per-file JSON outputs. Tools like jq can then query the extracted structures across files.

What are the limitations of automated concept extraction?▼

Extraction only captures what is stated or strongly implied in the source, so inferred relationships must be flagged with is_inferred. Confidence levels should reflect genuine uncertainty, and similar concepts must not be merged without explicit evidence they are identical.