Skip to content
Quarryv1.4

    Start free

    Changelog

    Releases in reverse order. A release marked breaking needs something from you before it will run the way the one before it did; the entry says what.

    Follow the release feed

    1.4.0Sep 10, 2026

    Saved searches now show live counts, and snippet extraction is faster across the board.

    Added

    • Live counts next to each saved search, updated as the index changes
    • quarry saved --watch to keep counts updating in the terminal
    • A [saved_searches] table in the settings file for default sort order
    • Keyboard shortcut to jump to a saved search from anywhere in the app

    Changed

    • Snippet extraction rewritten to skip files that cannot contribute a better snippet than one already found
    • Typical snippet extraction time cut by roughly 40% on large result sets
    • Snippet width in the app follows the window instead of a fixed 80 columns
    • The HTTP endpoint’s snippet parameter now short-circuits faster when set to false, matching CLI behavior
    • quarry status lists saved searches alongside indexed folders

    Fixed

    • Saved searches created in the CLI now appear immediately in the desktop app without a restart
    • Counts on saved searches no longer drift after a quarry reindex
    • The daemon no longer logs a spurious warning when no saved searches are configured

    1.3.2Jul 29, 2026

    A small patch release with five fixes and no new features.

    Fixed

    • The Neovim plugin no longer loses its connection after the daemon restarts
    • quarry config set now validates port numbers before writing them
    • Fixed a crash when indexing a folder containing a file with no extension and no readable text
    • The HTTP endpoint’s snippet=false parameter is now respected on the first request after a daemon restart
    • Corrected a rounding error in quarry status that under-reported index size on disk

    1.3.0Jun 16, 2026

    A Neovim plugin joins the VS Code plugin, and the snippet store was rewritten for lower memory use.

    Added

    • Neovim plugin, installable through the usual plugin managers
    • quarry saved command to list, run and delete saved searches from the terminal
    • index parameter on the HTTP endpoint, to search a single named index
    • Syntax highlighting for query terms in the desktop app’s search box

    Changed

    • Snippet store rewritten as a flat file with fixed-size records, replacing the previous log format
    • Memory use during indexing reduced by roughly a third on large folders
    • The VS Code plugin now reuses one connection to the daemon instead of opening one per search
    • quarry reindex runs in the background by default; pass --wait to block until it finishes

    Fixed

    • The daemon no longer restarts unnecessarily when the settings file is saved without changes
    • Saved searches with special characters in the name now display correctly in the app
    • Large PDF files no longer cause the indexer to hold files open longer than needed

    Removed

    • The undocumented --legacy-snippets flag, unused since 1.1.0

    1.2.0Apr 2, 2026

    A .quarryignore file adds per-folder exclusion rules, and the HTTP endpoint gains limit and snippet parameters.

    Added

    • .quarryignore files, read per folder using gitignore syntax
    • limit parameter on the HTTP endpoint, 1 to 200 hits
    • snippet parameter on the HTTP endpoint, set to false to skip snippet extraction
    • quarry status --json for scripting
    • A [ignore] table in the settings file for global exclude patterns

    Changed

    • Default HTTP limit set to 20, matching the CLI default
    • Ignore rules now apply during a rebuild, not only during the initial index
    • quarry search prints file paths relative to the indexed folder instead of absolute paths

    Fixed

    • Folders containing a .quarryignore with only comments no longer fail to index
    • Symlinked folders inside a watched directory are now indexed once, not repeatedly
    • The HTTP endpoint no longer returns a 500 when q is present but empty

    1.1.0Jan 20, 2026Breaking

    Index format v2 ships in this release; every existing index must be rebuilt with quarry reindex before searching again.

    Added

    • quarry reindex command to rebuild an index in place
    • Config option watch.debounce_ms to control how quickly file changes are picked up
    • quarry config get and quarry config set for reading and writing individual settings
    • Wildcard queries with * at the end of a term

    Changed

    • Index format moved to v2, with a smaller postings file
    • Snippet extraction now runs after ranking, not before, to save work on unranked hits
    • quarry status reports index size on disk alongside file counts
    • Default debounce for file watching raised from 200ms to 500ms

    Fixed

    • Searches issued while a large folder was still indexing no longer return partial results silently
    • PDF text layers with unusual encodings no longer stall the indexer
    • The daemon now releases its lock file cleanly on quarry daemon --stop

    1.0.0Oct 14, 2025

    First public release of Quarry, for macOS and Linux, with the desktop app, the CLI and the background daemon.

    Quarry is now available for macOS and Linux. This release covers the core loop: point the app or the CLI at a folder, build an index, and search it with a small, predictable query syntax. The desktop app, the quarry CLI and the background daemon all ship together and share the same index format, so behavior matches across all three. Free and Pro plans are both available from day one, with Free covering up to 20,000 files in a single index.

    • Index and search plain text, Markdown, PDF text layers and common source code
    • quarry index, quarry search and quarry status on the command line
    • Snippet previews in the desktop app
    • A settings file at ~/.config/quarry/config.toml