用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/kyre-99/Everything-is-Knowledge --skill wiki-query命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Ingest sources into the wiki using external LLM extraction (faster batch processing). Fully automated - fetches, extracts, and writes wiki pages in one command.
Initialize the Everything is Knowledge wiki. Creates directories, configures API keys, and sets up PDF parsing options. Run once to set up a new wiki.
Disambiguate entities and manage aliases for the knowledge wiki
基于 SOC 职业分类
正在显示 SKILL.md
| name | wiki-query |
| description | Query the knowledge wiki and get synthesized answers |
| allowed-tools | ["Read","Bash"] |
Query the wiki to find information. Uses Obsidian CLI for fast search and link graph traversal.
/wiki-query <question>
/wiki-query What is React?
/wiki-query How do attention mechanisms work?
/wiki-query What papers discuss agentic memory?
Use obsidian search to find relevant pages:
obsidian search query="<keywords>" format=json
Examples:
obsidian search query="RAG" format=json → returns files mentioning RAGobsidian search query="reinforcement learning" format=json → returns RL-related pagesThe search returns a JSON array of file paths like:
["wiki/entities/RAG.md", "wiki/entities/Reinforcement Learning.md"]
For each relevant entity, get its backlinks to understand context:
obsidian backlinks file="Entity Name" format=json
This shows which other pages reference this entity — useful for:
Read the content of relevant pages:
obsidian read path="wiki/entities/Entity.md"
Or use the Read tool for direct file access if Obsidian CLI is slow.
Combine information from the pages into a coherent answer.
## Answer
[Synthesized answer based on the wiki pages]
## Sources Consulted
- [[Entity Page]] — [what it contributed]
- [[Another Page]] — [what it contributed]
## Related Entities (via backlinks)
- [[RelatedEntity]] — linked from X pages
| Command | Purpose |
|---|---|
obsidian search query="X" | Full-text search across all wiki files |
obsidian search query="X" path=wiki/entities | Search only in entities folder |
obsidian backlinks file="X" | Find pages linking to entity X |
obsidian links file="X" | Find outgoing links from entity X |
obsidian read path="wiki/entities/X.md" | Read page content |
obsidian files folder=wiki/entities | List all entity pages |
wiki/
├── cache.md # Entity names (one per line)
├── entities/ # Entity pages with facts
├── raw/ # Source documents
└── log.md # Operation log
# EntityName
type: artifact/abstract/person/concept
## Facts
- [[Entity]] description... [[source-file]]
- [[Entity]] another fact... [[another-source]]
/wiki-ingest-llm."If Obsidian CLI fails or is not installed, use traditional approach:
wiki/cache.md for entity listgrep -l "keyword" wiki/entities/*.md