What is the difference between search and vsearch in this skill?

The 'search' command uses BM25 for keyword matching (finding exact terms), while 'vsearch' uses vector embeddings to find notes that are conceptually similar even if they don't share exact words.

Do I need to manually index my notes?

Yes, initial setup requires running 'qmd embed', and you should use 'qmd update' after making significant changes to your vault to keep the search index current.

When should I use the 'query' command instead of 'vsearch'?

Use 'query' for thorough exploration of a topic; it is higher quality but slower because it combines BM25 and vector search with LLM-based re-ranking.

Can I use this with any markdown folder?

Yes, as long as you register the folder as a collection using the qmd collection add command, it can index and search any directory containing markdown files.

Does this skill require cloud APIs for searching?

No, this skill uses the qmd engine for local semantic search, meaning your data stays on your machine and search operations are performed entirely offline.

Semantic Vault Search

Name: Semantic Vault Search
Author: JSai23

byJSai23

•

内容管理

Enables high-performance local semantic and keyword search for markdown note collections using the qmd engine.

Semantic Vault Search integrates the qmd local search engine into Claude, allowing for sophisticated retrieval across markdown note vaults. It supports traditional keyword search (BM25), vector-based similarity searches for conceptual discovery, and high-quality hybrid queries with LLM re-ranking. Designed for privacy-conscious users, all search operations and embeddings are handled locally, making it an essential tool for managing large personal knowledge bases, technical documentation, or project notes without relying on external cloud APIs.

主要功能

01Fast BM25 keyword search for specific term matching and lookups

02Hybrid search with LLM re-ranking for maximum retrieval accuracy

031 GitHub stars

04Local vector search for finding conceptually related notes without keyword overlap

05Automated vault indexing and embedding management via the qmd utility

06Multi-format output support including JSON, CSV, and Markdown for programmatic use

使用场景

01Determining the most appropriate filing destination for new content based on existing context

02Performing thorough research on specific topics within a local vault using hybrid search

03Discovering semantic connections between notes across different folders in a knowledge base

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jsai23/claude-tooling vault-sem-search-a

For use in Claude.ai and ChatGPT

Download Skill