Commands

Command	What it does
classify	Tag photos as photo/screenshot/document/meme
config	Show or change defaults
dedupe	List duplicate copies you could delete
embed	Prepare photos for search (one-time, resumable)
export	Write labels (faces, location, ratings) to portable .xmp sidecars
faces	Detect and group faces
fix-dates	Set each file’s date from its EXIF shoot date
gallery	Browse the library in a local web UI: files, duplicates, dates, people, map, events
import	Bring photos in from Google Takeout, Apple Photos, or Lightroom
locations	Group photos by where they were taken
mark	Star rating, like, keep/reject, label your media files
mcp	Expose search to AI agents
pipeline	Bring the whole library current in one command (scan through locations)
prune	Remove database entries for files that no longer exist
scan	Read a folder and record what’s in it. Run this first.
search	Find photos by description, example image, person, category, or place
stats	Library totals and what has run recently
status	Is the library up to date, what to run next, and is the watcher alive
tag	Add or remove free-form tags; filter with search --tag
watch	Background loop keeping everything current

Every command also takes --help.

Common options

Most commands share these:

Option	Effect
--library <dir>	Operate on the library at <dir> instead of the invocation directory
--silent	Suppress progress output on stderr
--json	Print one JSON object on stdout instead of human-readable text
--model <id>	Use a specific search model (embed, search, classify, gallery, mcp)

What needs what

To run	You first need
dedupe, gallery, fix-dates, prune, stats, locations	scan
scan, when coming from another tool	import first, so dates are fixed before they are recorded
search (text or image)	embed
search --person	faces, then naming via gallery
search --category	classify (which needs embed)
search --location	GPS data in your photos
search --date / --after / --before	nothing beyond scan

Workflows has the full dependency map, what to run afterwards (prune is the one people forget), recipes for the common jobs, and rough costs for the long ones.