qmd

name	qmd
description	Search the notes mind using QMD hybrid search. Use when looking up notes, finding related documents, searching for decisions, people, initiatives, or any mind content. Triggers on search, find, look up, "what do I know about", "where is", recall.

QMD — Mind Search

Local hybrid search engine indexing this mind. BM25 keyword + vector semantic + LLM reranking.

Status

!qmd status 2>/dev/null || echo "Not installed: npm install -g @tobilu/qmd"

Collections

Collection	Content
domains	Recurring areas: team notes, people notes, operational concerns
initiatives	Active projects and their status, decisions, scope
expertise	Learning notes, technical references, patterns, frameworks
inbox	Unprocessed captures awaiting triage
working-memory	Agent working memory: observations, conventions, preferences

CLI Usage

qmd search "keywords"               # BM25 only — fast, exact matches
qmd vsearch "natural language"       # Vector semantic search
qmd query "complex question"         # Full hybrid + reranking (best quality, slowest)
qmd get "#docid"                     # Retrieve by docid from search results
qmd multi-get "journals/2026-*.md"   # Batch retrieve by glob

Structured Queries (advanced)

qmd query $'lex: sprint review accountability\nvec: how does the team handle check-ins and progress tracking'

Query Types

Type	Method	When to use
lex	BM25 keyword	Exact names, IDs, specific phrases — "sprint-review", "ADO 2049967", "project-alpha"
vec	Vector semantic	Conceptual questions — "what coaching frameworks does the team use"
hyde	Hypothetical doc	You know what the answer looks like — write 50-100 words of expected content

Writing Good Queries

lex (keyword)

• 2-5 terms, no filler words
• Exact phrase: "connection pool" (quoted)
• Exclude terms: performance -sports (minus prefix)
• Proper nouns work best here: person names, project names, specific terms

vec (semantic)

• Full natural language question
• Be specific: "what are the gates for private preview"
• Include context: "what decisions were made about the prod deployment sequence"

hyde (hypothetical document)

• Write 50-100 words of what the answer looks like
• Use the vocabulary you expect in the result

Combining Strategies

Goal	Approach
Know exact terms	qmd search (BM25 only, instant)
Don't know vocabulary	qmd vsearch or qmd query
Best recall on complex topic	qmd query with lex + vec lines
Quick daily use	qmd search for most lookups

First query line gets 2x weight in fusion — put your best guess first.

Collection Filtering

qmd search "API review" -c initiatives    # Search only initiatives
qmd vsearch "coaching" -c domains         # Search only domains

Omit -c to search all collections.

Output Formats

qmd search "term" --json              # Structured JSON for processing
qmd search "term" --all --files       # File paths only, all matches
qmd search "term" --min-score 0.5     # Filter by relevance threshold