explaingit

bugjay/weread-skill

0Audience · writerComplexity · 2/5ActiveLicenseSetup · easy

TLDR

Claude Code Skill that converts WeRead ebook highlights and notes into atomic knowledge cards, concept maps, and review questions for a personal RAG-ready library.

Mindmap

mindmap
  root((weread-skill))
    Inputs
      WeRead Markdown export
      User comments
      Book metadata
    Outputs
      Atomic cards
      Concept map
      Review questions
      Source evidence file
    Use Cases
      Build personal knowledge base
      Feed cards into RAG
      Prep book review notes
    Tech Stack
      Claude Code
      Markdown
      JSONL

Things people build with this

USE CASE 1

Turn WeRead highlights into atomic knowledge cards for Obsidian or Notion

USE CASE 2

Generate review questions from a finished book

USE CASE 3

Build RAG-ready chunks with chapter and location metadata from a book

Tech stack

Claude CodeMarkdownJSONL

Getting it running

Difficulty · easy Time to first run · 5min

Clone into the Claude Code skills directory and the skill auto-activates on Chinese trigger phrases or pasted WeRead notes.

MIT license lets anyone use, modify, and redistribute the code as long as the copyright notice is kept.

In plain English

This repository is a Claude Code Skill, meaning a small package of instructions that the Claude Code tool can load to perform a specific task. The README is written in Chinese. The Skill takes reading notes exported from WeRead, a popular Chinese ebook app run by Tencent, and turns them into a personal knowledge base. WeRead lets readers save highlights, thoughts, chapters, positions, and dates from the books they read. The Skill ingests that exported Markdown or TXT file and produces a structured set of outputs: a source evidence file that keeps the author's original text and the user's comments clearly separated, a candidate notes file of high value points, atomic knowledge cards with one idea per card stored in both Markdown and JSONL form, a concept relationship map showing how ideas support, contradict, or extend each other, a set of review questions, a file of reusable application assets like writing prompts, and an index for the whole book. The README is firm about a few rules. The three voices, author original text, user comments, and AI processing, must always stay distinct. Every card must cite its book, chapter, and location. AI is told to never invent theories, chapters, or page numbers that are not in the source. The output is also designed to be ready for retrieval augmented generation systems, where each atomic card is one chunk with metadata for the book, author, chapter, location, and card type. Installation is by cloning the repo into the Claude Code skills directory, after which the Skill auto activates on Chinese trigger phrases like organize reading notes or generate atomic cards, or when a user pastes WeRead exported notes. The README also lists adaptation strategies for theory books, method books, business books, and other categories. License is MIT.

Copy-paste prompts

Prompt 1
Install weread-skill into my Claude Code skills directory and trigger it on a pasted WeRead export
Prompt 2
Run weread-skill on this exported Markdown and produce atomic cards plus a concept map
Prompt 3
Adapt the skill rules for a business book versus a theory book
Prompt 4
Show me the JSONL schema each atomic card produces and how to load it into a vector store
Open on GitHub → Explain another repo

Generated 2026-05-22 · Model: sonnet-4-6 · Verify against the repo before relying on details.