LLM Wiki Knowledge Base
Paste the following prompt into your AI chat to install this skill:
Please install @user_15292d5a/yjkj-llm-wiki-q into your AI assistant using https://skillhub.cn/install/skillhub.md.
About this skill
Problem
RAG re-retrieves raw documents on every query, so knowledge does not accumulate and AI mistakes lack a stable correction surface. llm-wiki-q treats knowledge as a persistent wiki: the AI compiles sources, and humans correct them through structured audit feedback.
How It Works
Each session starts from CLAUDE.md and wiki/index.md, then uses five operations:
- compile: split oversized pages, merge near-duplicates, and rebuild the index
- ingest: save articles, papers, notes, or large-file pointers under raw/, then update summaries, concepts, and entities
- query: answer from the wiki, cite pages with [[Page]], and promote durable answers
- lint: check dead links, orphan pages, missing index entries, and audit format
- audit: process audit/ feedback by accepting, partially accepting, rejecting, or deferring, while preserving resolution history
Pages target 400–1200 words; complex topics split into subpages. Diagrams use mermaid, formulas use KaTeX. Large files become raw/refs/ pointers so the repo stays portable.
Boundaries
Best for research notes, personal wikis, team knowledge bases, and reading companions. If sources conflict or audit feedback piles up, the AI can still be wrong. It assumes humans stay in the loop rather than fully automatic encyclopedia building.
Use Cases
- Read papers and blog posts over weeks, then compile them into cross-linked concept pages and maintain the index.
- Ingest meeting notes and Slack discussions, generate summaries, and correct issues through web audit feedback.
- Split oversized personal notes, merge near-duplicates, and check dead links, orphan pages, and missing index entries.
- File book chapters as you read, then promote durable answers into concept pages.
Best For
- Researchers tracking a field over weeks who want papers, articles, and questions compiled into a continuously updated wiki.
- Engineering or product leads who need team discussions, docs, and meeting notes into a queryable knowledge base with correction history.
- Technical writers or learners maintaining personal notes who want the AI to split, link, and organize content while they review and audit.
- Obsidian or local-web users who want cross-references, Mermaid diagrams, formulas, and audit feedback in one wiki workflow.
Related Skills
Search Huawei Cloud official docs and product pages to find ECS, OBS, RDS, CCE product specs, parameters, documentation, and API references without login.
OCR-based recognition for movie, train, flight, and event tickets in images or PDFs, extracting key fields into Markdown or JSON reports.
Turns notes, research, and meeting summaries into actionable next moves, plans, decisions, experiments, and decision-changing gaps.
A local wiki knowledge base manager that compiles raw documents into sourced, indexed Markdown pages with wikilinks, query support, and health checks.