Introduction

Quarry is a local-first index and search tool for notes, documents and code. Version 1.4, released September 2026, runs on macOS and Linux. It indexes the folders you point it at and answers searches from that index in place. Nothing is uploaded and nothing is synced.

This page covers what Quarry does, what it does not do, and how the three parts, the desktop app, the CLI and the background daemon, work together. Start with Install if you already know you want it running, or keep reading for the shape of the thing first.

What gets indexed

Quarry watches only the directories you choose. There is no background crawl of your home directory, your whole disk, or any folder you have not explicitly added. Point it at a notes vault, a documents folder, a git checkout, or all three. Each becomes part of an index, or its own index, depending on how you set it up.

Supported content includes plain text, Markdown, PDF text layers, and source code across the common languages. Binary files without extractable text are skipped. You control what is included or excluded per folder with include and exclude rules.

The three parts

The desktop app is where most people spend their time. It shows index status, lets you add and remove watched folders, runs searches with a snippet preview, and holds the settings that also live in the settings file.

The CLI, invoked as quarry, does everything the app does and a few things it does not. It is the same binary the app calls internally, so behavior matches exactly.

quarry index ~/notes
quarry search "invoice AND q3"
quarry status

The daemon is the background process that actually holds the index open, watches for file changes, and, if you enable it, answers searches over the loopback HTTP endpoint. The app starts the daemon for you. On a headless Linux box you can run it directly:

quarry daemon --serve

See Command line for the full command reference.

How search works, briefly

Each indexed folder is scanned once, then watched for changes. New and modified files are picked up within a few seconds, according to your file system’s change notifications, not by polling on a timer. A search reads the query syntax you type, ranks matches with a small set of rules, and returns results with a text snippet around the match.

Frequent searches can be turned into saved searches, which show up in both the app and the CLI without retyping the query.

What Quarry is not

Quarry is not a sync tool. It does not copy files anywhere, and it does not keep multiple machines in agreement. If you index the same folder on two machines, you get two independent indexes.

Quarry is not a cloud search service. The HTTP endpoint it can expose, Pro only, binds to 127.0.0.1 and nothing else. There is no server component, no account system beyond a license key, and no telemetry.

Quarry is not a full-text database with a query language for building applications. The query syntax is built for humans searching their own files quickly, not for constructing arbitrary structured queries.

Free, Pro and Team

The Free plan indexes up to 20,000 files in a single index, with three saved searches. That is enough for most personal notes vaults. Pro removes those limits, adds unlimited indexes, and turns on the HTTP endpoint and the editor plugins. Team adds shared rule files across a group’s machines. Details and current pricing are on /pricing.

Where to go next

1. Install Quarry. See Install for macOS and Linux instructions.
2. Index a folder. Index your first folder walks through the first quarry index run and what to expect.
3. Learn what Quarry writes to disk. Project structure covers the config file, the data directory and what is safe to delete.
4. Tune what gets indexed. Include and exclude rules explains .quarryignore and config-level filters.

Release notes for each version are on /changelog.