What is the 'intent' feature for?

Intent disambiguation allows you to provide a steering signal to the search engine when a keyword might have multiple meanings, ensuring the AI retrieves the relevant context.

It is a specialized skill for Claude Code that enables advanced searching of project-specific markdown documentation using both semantic and keyword-based methods.

How does hybrid search work in this skill?

The skill combines BM25 keyword matching for exact terms with vector embeddings for conceptual meaning, scoring and fusing results to provide higher accuracy than standard search.

Do I need to configure anything to use QMD search?

Yes, your project must have a .claude/qmd.json configuration file and the qmd MCP server must be installed and running.

What happens if the semantic search finds no results?

The skill includes robust error handling that falls back to CLI queries and eventually standard Glob/Grep searches to ensure information is found even if indexing is stale.

QMD Semantic Search

Name: QMD Semantic Search
Author: aviflombaum

byaviflombaum

•

学习与文档

Searches project markdown documentation using advanced semantic, lexical, and hybrid query techniques for high-precision information retrieval.

The QMD Search skill empowers Claude to navigate and retrieve information from your project's internal documentation with sophisticated precision. By leveraging the qmd MCP toolset, it performs hybrid searches combining BM25 keyword matching, vector-based semantic understanding, and hypothetical document embeddings (HyDE). This allows developers to find specific implementation details, project plans, or architectural decisions even when the exact terminology isn't known, ensuring Claude always has the right context to provide accurate assistance.

主要功能

01Automated configuration reading from .claude/qmd.json

02Hybrid search combining keyword (lexical) and vector (semantic) methods

03High-speed MCP integration with fallback to CLI and Grep

0436 GitHub stars

05HyDE support for complex queries via hypothetical document generation

06Intent disambiguation to filter results for ambiguous terms

使用场景

01Finding relevant project plans or feature requirements before starting a task

02Retrieving specific architectural patterns or coding standards from project docs

03Exploring large markdown knowledge bases to understand unfamiliar codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add aviflombaum/claude-code-in-avinyc search

For use in Claude.ai and ChatGPT

Download Skill