SPA3K

Quantum Reading Skill

Name: Quantum Reading Skill
Author: SPA3K

AI community

SOURCE https://github.com/SPA3K/readless-skill

Description

Read Less, Understand More - Lightweight intelligent text reader for AI agents with 30%+ token optimization

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

⚛️ 量子阅读 Quantum Reading

40% of the text. 0% of the value. We skip it.

**Like quantum tunneling — bypass the noise, reach the information core instantly.**

[English](README.md) | [中文](README_CN.md) | [Live Demo](web/index.html)

🔬 Experiment Results

Quantum Reading applied to three real-world Chinese text samples:

Input	Type	Original Tokens	After QR	Saved	💰 Cost Saved*
GAPS 论文	Physics paper (arXiv)	2,008	1,388	31%	$0.0093
微信推送文	WeChat tech article	2,336	1,477	37%	$0.0129
表白小作文	Love confession letter	1,189	681	43%	$0.0076

*Based on Claude Opus 4.7 input pricing ($15/MTok). Savings scale with document length — longer docs yield higher reduction.

🖥️ Web Demo

Try it now — paste any text and watch Quantum Reading highlight what matters.

**[→ Open Demo](web/index.html)**

The demo visualizes every rule from the skill: summary markers, structural headings, data patterns, logic connectors, and paragraph first/last sentence priority. High-value sentences appear in **bold white**, low-value text is grayed out.

![GAPS Paper — 31% saved](screenshots/article.jpg)

*GAPS physics paper (arXiv) — 2,008 → 1,388 tokens, 31% saved*

![WeChat Article — 37% saved](screenshots/wechat.jpg)

*WeChat tech article (ChatGPT Images 2.0) — 2,336 → 1,477 tokens, 37% saved*

![Love Letter — 43% saved](screenshots/balabala.jpg)

*Love confession letter — 1,189 → 681 tokens, 43% saved*

Overview

Quantum Reading is an AI Agent skill that teaches AI to **skim documents like humans do** — not word-by-word, but through intelligent scanning strategies.

Through **8 human reading techniques** and **Subagent isolation**, it saves ~40% tokens on typical texts while maintaining 98%+ information accuracy.

Works with Claude Code, C