QMD Expert Workflow

This skill outlines the most effective workflow for AI agents to query, extract, and fetch relevant context from a knowledge base using the qmd CLI tool.

Because qmd chunks documents under the hood, standard search commands often return specific, highly relevant chunks (identified by a #hash) rather than entire files. To ensure you don't lose the pointer to the most relevant information, follow this robust workflow.

1. Understand QMD Capabilities

If you are ever unsure about the syntax, query types (lex, vec, hyde, expand), or available flags, you should start by running:

qmd skill show

This will output the embedded QMD documentation detailing how to write good queries, use intent for disambiguation, and combine search types for the best recall.

2. Formulating Queries

Identify a list of diverse queries to cast a wide net over the knowledge base. Consider mixing query types based on the user's need: