1.2.1

Minor Bug fixes

1.2.0

What's new

• Snippet preview: see the best-matching sentence from each note directly in the inline suggest popup and Related Notes panel
• Uses a 1-bit quantized sentence index (32× smaller than Float32) — same storage footprint as the existing note index
• Three independent toggles: enable snippets (requires re-index), show in popup, show in panel
• When disabled: zero performance overhead, existing index unchanged

1.0.4

What it does

Suggests [[note links]] as you type — combining instant lexical matching
with semantic search from a local embedding model. A sidebar panel shows
the most related notes for whatever you have open.

Requirements

• Ollama with ollama pull bge-m3, or
• LM Studio with any embedding model loaded, or
• An OpenAI API key (set in plugin settings)

Install

Download the three files below, place them in
.obsidian/plugins/semantic-backlinks/ and enable the plugin.
Or use BRAT → Brightwav3/semantic-backlinks.

Features

• Inline popup with lexical + semantic results while you type
• Related Notes sidebar panel
• Chunked indexing — long notes aren't truncated
• Binary embedding storage (~7× smaller than JSON)
• Auto-reindex after edits, live progress on first run
• Excluded folders, adjustable thresholds, mobile via Tailscale
• All note content stays on your device (unless you choose OpenAI)