Include and exclude rules

By default Quarry indexes every text-extractable file under a watched folder. Most folders have things you do not want in the index: build output, dependency directories, drafts. Two mechanisms narrow that down, and they can be used together.

.quarryignore files

Place a file named .quarryignore next to the content it applies to. The syntax matches .gitignore: one pattern per line, # starts a comment, and a leading ! re-includes a path an earlier pattern excluded.

# build output
dist/
node_modules/

# temp files
*.tmp
*.swp

# but keep this one build artifact
!dist/manifest.json

Quarry reads .quarryignore files at index time and again whenever the daemon detects a change to one. A pattern applies to the directory the file lives in and everything below it, the same scoping rules as git.

Pattern syntax

Order matters. Later lines can override earlier ones, but only within the same file; a pattern in a parent directory’s .quarryignore cannot be un-excluded by a child directory’s file unless the child repeats the negation.

Config-level rules

The same effect can be set in the settings file, under [index.<name>]. This is useful when you do not want to add a file to the folder itself, or when you are scripting index setup.

[index.notes]
path = "/Users/dana/notes"
exclude = ["*.tmp", "drafts/**", "archive/2019/**"]
include = ["archive/2019/keep-this.md"]

exclude and include here use the same gitignore-style syntax as .quarryignore. An include entry re-adds a path that an exclude pattern would otherwise drop, the same way a ! line does in a .quarryignore file.

How the two combine

When both are present, config-level rules are applied first, then any .quarryignore files found while walking the folder. A file excluded by either mechanism is excluded; there is no way for a .quarryignore file to override a config-level exclude, only to add more excludes of its own or negate its own.

Checking what will be excluded

Before running a full index, preview which files a rule set would drop:

quarry index ~/notes --dry-run

would index:   1,204 files
would exclude:   318 files (see --list-excluded)

Add --list-excluded to print every excluded path, useful for confirming a broad pattern like drafts/** is not catching more than intended.

Rules for code folders

For a folder that is also a git repository, a common pattern is to reuse the existing .gitignore conventions by copying the relevant lines into .quarryignore, since Quarry does not read .gitignore itself. This keeps the two files independent: something you want git to ignore is not automatically something you want left out of search, and vice versa. A node_modules/ directory is a good candidate for both. A secrets.env file you keep out of git for a different reason might still be worth indexing locally, since Quarry’s index never leaves your machine.

Shared rule files

Team plans can share a .quarryignore-style rule set across every machine on the team, stored centrally and applied automatically to matching folders on each member’s daemon. This is a Team-only feature; see /pricing for how it differs from Pro. On Free and Pro, rule files are always local to the machine they are written on.

Re-applying rules to an existing index

Adding or editing exclude rules does not retroactively remove already-indexed files. Run:

quarry reindex notes

This rebuilds the index from the source files using the current rules. See Index your first folder for what a rebuild does, and Project structure for what gets rewritten on disk.