# Smart Connections

> Semantic search and related-note surfacing inside Obsidian, with a local embedding model

- Category: AI Productivity & Writing / Notes & Knowledge Bases
- Language: JavaScript
- Licence: Smart Plugins License Agreement (source-available)
- Repository: https://github.com/brianpetro/obsidian-smart-connections
- Website: https://smartconnections.app
- Canonical page: https://ai-tldr.dev/tools/smart-connections/

## In short

Smart Connections is an Obsidian plugin for the problem every large vault eventually has: the note you need exists, and you have forgotten it. It sits in the Notes & Knowledge Bases category of the AI Productivity & Writing tools and is open-source, written in JavaScript and licensed Smart Plugins License Agreement (source-available).

## Overview

Smart Connections is an Obsidian plugin for the problem every large vault eventually has: the note you need exists, and you have forgotten it. It embeds your vault with a local model and then continuously shows the notes and excerpts most semantically related to whatever you are writing right now, in a Connections view you can read as a ranked list or a mini graph and drag links out of directly.

The design choice that sets it apart from most AI note plugins is that it does not phone home. The plugin ships with a local embedding model that starts indexing as soon as you enable it — no extra apps, no CLI tools, no API key — and works offline and on mobile. The author leans on a deliberately small dependency surface, arguing the codebase can be audited several times faster than comparable AI plugins.

Alongside the Connections view there is a Lookup view for running semantic searches across the vault on demand, and a settings layer (the Smart Environment) for tuning indexing, filters, ranking and UI defaults on large or specialised vaults. The core is source-available under the project's own Smart Plugins License Agreement, with paid Pro surfaces layered on top — inline connections that suggest notes while you type in the editor, and an Obsidian Bases integration.

## What it does

- Connections view: semantically related notes and excerpts for the current note, as a ranked list or mini graph, with links you can drag straight into your writing
- Lookup view for on-demand semantic search across the whole vault
- A bundled local embedding model — zero setup, no API key, works offline and on mobile
- Smart Environment settings for indexing, filters, ranking and UI defaults, aimed at large or specialised vaults
- Ultra-lightweight bundle with minimal third-party dependencies, written to be auditable
- Pro surfaces on top of the free core: inline connections while writing, and an Obsidian Bases integration

## Getting started

The plugin is designed to work with no configuration — the README's own three-step plan is install, keep writing, open the view.

1. Install from Obsidian Community plugins

   Find Smart Connections in the Obsidian Community plugins directory and enable it. No extra apps, no CLI tools, no API key.

```text
Settings → Community plugins → Browse → "Smart Connections" → Install → Enable
```

2. Let it index

   The bundled local model starts creating embeddings for your vault immediately; a notice tells you when embedding is complete. Just keep writing while it works.

3. Open the Connections view

   From the left ribbon (the Connections icon) or the command palette. Related notes appear for whatever note you are in, and you can drag links from the list into your text.

```text
Command palette → "Open: Connection view"
```

4. Search on demand with Lookup

   Use the Lookup view when you want to ask the vault a question directly rather than see what relates to the current note.

5. Tune it for a big vault

   The Smart Environment settings control indexing, filters, ranking and UI defaults; the Getting Started guide is reachable from the plugin's settings pane.

## When to use it

- Surface forgotten prior work while drafting, so research you already did resurfaces instead of being repeated
- Build links and structure in a large vault semantically rather than by remembering exact titles or tags
- Run semantic search over personal or sensitive notes on a machine that never sends them to a hosted model
- Keep the same related-notes workflow on mobile, where the local model still runs