Semantic Layer [[semantic: __Namespace]]
Semantic search for QMDC workspaces.
Overview
QMDC Semantic provides semantic search capabilities for QMDC workspaces:
- Hybrid Search: Combines keyword search (FTS5), trigram substring matching, and dense vector search
- Graph Walk: Expands results through explicit and inferred edges
- Inferred Edges: Discovers semantic relationships between objects
- Incremental Indexing: Hash-based updates for efficiency
Quick Start
Installation
# From the qmdc repo root
uv pip install -e ./qmdc-py -e ./qmdc-semantic
Index a Workspace
qmdc-semantic index /path/to/workspace
Creates .qmdc-semantic/embeddings.db with:
- Chunk embeddings
- FTS5 index
- Inferred edges
Search
# Text query
qmdc-semantic search /path/to/workspace "how to test LSP"
# Impact scan (query from file)
qmdc-semantic search /path/to/workspace --from-file task.qmd.md -k 20
Configuration
Create .qmdc-semantic/config.yaml in workspace or ~/.qmdc-semantic/config.yaml globally.
See Semantic Configuration for details.
Namespace Contents
- Semantic CLI Commands — CLI reference
- Semantic Algorithms — Algorithm documentation
- Semantic Configuration — Configuration guide
- Semantic Storage Schema — Storage schema
- Semantic Testing — Testing guide